Background: Birth weight is an important determinant of neonatal and child health outcome. For instance, evidence has shown
that low birth weight (LBW) has a negative impact on the baby’s growth, cognitive development, and on neuro-motor development
and immune function. Underweight babies are prone to increased risk of infections and stunting. LBW remains prevalent worldwide
and is more pronounced in low- and middle-income countries. Several factors including mothers’ socio-economic characteristics,
maternal health behaviors and maternal and pregnancy health conditions determine birth weight outcome. This study sought to
determine the prevalence and investigate determinants of LBW among Burundian women of reproductive age.
Methods: This study used data extracted from the 2016-2017 National Demographic and Health Survey (DHS) conducted on
7047 women who reported a live birth history in the five years preceding the survey and whose birth weight was recorded at
childbirth. The study used linear regression to explore socio-economic, maternal, and pregnancy related factors that determine
birth weight and further employed a logistic model to unpack factors with higher likelihood of LBW.
Results: Of 7,047 babies born between 2012 and 2017, 660 (10%)
were underweight. Findings suggested that birth weight
decreases with older women’s age, multiple pregnancies (twin or
triplet), and female babies. Conversely, birth weight increases
with a diabetes condition, wealthier quintiles, and higher party orders.
Results from the linear regression were supported by those
implemented in the discrete model. In fact, higher parity orders and
wealthier women were more likely to deliver normal weight
babies. High blood pressure, smoking, multiple pregnancies, and female
child’s sex were negative predictors of normal birth weight.
For instance, twin babies were twice more likely to be underweight
compared to single pregnancies.
Conclusion: This study unpacked high prevalence of LBW in Burundi and further highlighted areas of improvement to deliver
on global neonatal and child health targets.
Results: From this work could be used to implement targeted interventions to reduce poverty, tackle chronic conditions in
pregnancy, and reduce tobacco use among pregnant women as the above predicted LBW. Other interventions include modern
contraception through health educational programs.
Keywords: Low birth weight; Determinants; Burundi
Introduction
Normal birth weight, defined as live-born neonates weighing
from 2500 g to 4000 g at birth, is an important determinant of
better neonatal and child health. Evidence has established the
impact of low birth weight (LBW) on increased neonatal deaths
and child stunting [1,2]. Furthermore, low-birth babies experience
perinatal growth failure, reduced cognitive and neuro-motor
functioning, and poor school performance [3, 4]. There exists a
widening negative correlation between birth weight and children’s
better health with extreme low-birth-weight babies experiencing
major health conditions [4, 5]. Most importantly, underweight
babies who survive tend to have impaired immune function and
increased risk of disease; they are likely to remain undernourished,
with reduced muscle strength, cognitive abilities, and intelligence
quotient (IQ) throughout their lives [6]. Despite efforts to improve
pregnancy experience and combat LBW, many countries still report
high numbers of underweight babies until today. In 2012, the World
Health Assembly (WHA) resolution set the goal to achieve 30%
reduction in the number of LBW newborns by 2025 [7]. This target
was further reemphasized by the Sustainable Development Goals
(SDGs) agenda by setting an aim to “end all forms of malnutrition”
including among pregnant women to “reduce neonatal mortality to
at least as low as 12 per 1,000 live births and under-5 mortality
to at least as low as 25 per 1,000 live births” [8]. Global efforts did
not yield expected results as some countries continue to experience
high rates of LBW.
Of about 21 million low birth babies – representing nearly 15%
of all live-births worldwide – 19 million of too small babies are
from Africa and Asia (excluding Japan). Altogether, more developed
regions namely Northern America, Europe, Japan and Australia,
and New Zealand account for less than 10% of low births [5].
According to available data, the whole sub-Saharan Africa excluding
Rwanda suffers from LBW. Countries with higher LBW rates (15 to
20% of all live births) include Angola, Benin, Burundi, Botswana,
Cote d’Ivoire, Guinea Equatorial, Madagascar, Namibia, Togo, and
Western Sahara. Guinea Bissau exceeds 25% of LBW rate and is
among the highest worldwide [5]. Despite Burundi having declined
LBW from 17.4% [CI: 11.9–22.9] in the year 2000 to 15.1% [10.9–
19.4] in 2015; the country remains above the global average and
is further top-ranked LBW rates [5, 9]. To better tackle LBW and
deliver on maternal and children global targets, countries need
to invest into evidence-based interventions that have been found
to undermine childbearing and fraught with child growth among
pregnant women.
An amounting evidence has established a causal link between a
woman’s characteristics, maternal and pregnancy health conditions,
and maternal health behaviors with birth weight outcome. For
instance, WHO developed a framework that explains causal
pathways for an increased risk of LBW. The framework comprises
distal and proximal or immediate factors leading to small babies.
Among distal determinants of LBW include
a) maternal characteristics such as extreme age, multiple
parity, poor birth spacing, and wealth index.
b) maternal health conditions namely chronic diseases
which have been found to increase maternal risk (i.e. high blood
pressure and diabetes).
c) maternal malnutrition characterized by anemia and
extreme maternal weight.
d) and other risk behavioral factors such as increased alcohol and tobacco consumption [5]. During the course of the pregnancy,
LBW can result from a premature birth (a birth occurring before 37
weeks of pregnancy) and/or the growth faltering in the mother’s
womb [5]. Other researchers of whom Alfred Kwesi Manyeh in
Ghana [10] and Getaneh Baye Mulu in Ethiopia [11] found similar
evidence. They both established the effect of the mother’s age,
wealth, parity, gestational hypertension, maternal height, antenatal
care (ANC), mother’s education attainment and the child’s sex
on birth weight [10, 11]. Furthermore, a study conducted on 10
developing countries incriminated the place of the woman’s age,
ANC, literacy level, body max index, and wealth on babies weight
at birth [12]. Despite the topic being of national focus today, little
has been done to explore factors leading to LBW in Burundi. The
aim of this study was to determine the prevalence and investigate
determinants of LBW in Burundi. Results of this study inform the
design of maternal and neonatal policies with an aim to deliver on
global and national targets by 2030.
Methods
Source of data
This study is a secondary data analysis using the Burundi
Demographic and Health Surveys (DHS) 2017 datasets. To better
understand predictors of birth weight, the study used women’s
individual recorde dataset. This study used a sample of 7,047
women who reported a birth history during five years prior to the
survey and whose information on birth weight was included in the
dataset.
Outcome and explanatory variables
This study used “birth weight” as the dependent variable which
was considered as continuous first to allow a linear regression
analysis and again as dichotomous to enable the discrete model.
We based on WHO guidelines to define cut-offs of the dichotomous
“birth weight” outcome [13]. A dummy variable was generated
taking value 0 for babies weighing less than 2,500 g at birth and
value 1, otherwise. Selection of independent variables was informed
by the literature search and by the understating of local context.
We included individual woman’s characteristics, behaviors and underlying
health conditions as well as factors related to pregnancy
health. With an aim to better grasp the effect of coefficients on birth
weight, all explanatory variables were categorized as summarized
in Table 1.
Table 1: Socio-economic and demographic characteristics.
Key: X+SD: Mean + Standard deviation.
Data management and models specification
In the first instance, owing to DHS study design which used
multiple sampling stages, the dataset was survey set before
analysis. In the second stage, we constructed the linear and logistic
models as specified below:
In the above linear estimation, Yi is the outcome variable (i.e.
birth weight). The model includes an intercept and a random
error term. Independent variables are represented by a vector of
covariates Xi and B1 captures the magnitude of change in birth
weight corresponding to a unit change in explanatory variables.
Significance of linear coefficients was ascertained based on p-value
at α = 0.05.
Equation 2 is the logistic specification model. In this model, the
dependent variable is the log odds that a woman i delivers a normal
weight baby (i.e. a baby weighing at least 2,500 g) relative to those
giving birth to underweight babies (i.e. babies with less than 2,500
g). 0 β captures fixed effects and 1 β detects random effects on
the probabilities of giving birth to normal weight babies. The vector
of covariates Xi includes independent variables described in Table
1. For the logistic model, significance of explanatory factors was
determined based on a corresponding 95% confidence interval that
does not contain value 1.
Results
Results of this study are sectioned into three main subheadings.
In the first instance, we describe socio-demographic characteristics
of study participants. In the second time, we summarize findings
from the linear model of socio-economic and maternal health
factors on birth weight. In the final stage, we present results of the
logistic model on the probability of low versus normal birth weight.
Socio-economic characteristics
Table 1 summarizes socio-economic characteristics of
participants. Of 7047 women of reproductive age, half were aged
between 25 and 34 years old. Overall, majority of women did not
achieve university education. Further, the prevalence of tobacco
and alcohol consumption as well as the prevalence of health
conditions with evidence to complicate pregnancy or childbirth
was considerably low. Results also showed that mothers delayed
in attending ANC as only about 7% attended their first ANC in the
first trimester and only about 16% achieved recommended four or
more ANC visits during the course of the pregnancy. More than half
of surveyed women were married, and majority had a parity of two
to three. Surveyed women were evenly scattered across the wealth
quintiles. Low birth weight babies represented the vast minority
(less than 10%) and male babies constituted a slightly higher
proportion.
Determinants of birth weight
As can be viewed in Table 2 below, birth weight significantly
decreases with older women’s age and increases with a diabetes
condition. Also, there was evidence of a sharp increase in birth
weight with wealthier quintiles. For instance, women who belong
to the richer and richest quantiles gave birth to babies with
almost 130 grams and 200 grams more; respectively. Moreover,
there was a significant increase of birth weight with party.
Mothers who had four to five and six and more parity gave birth
to babies with 253grams and 320 grams more respectively. Most
importantly, diabetic women gave birth to babies with 750 grams
more compared to women without diabetes condition. The effect
of a woman’s age on birth weight is linearly negative with a 70 g
decrease for women above the age of 35 years. In contrary to the
above, multiple pregnancies and female babies are associated with
low birth weights. The decrease in birth weight is more evident for
triplet pregnancies; reaching nearly 1800 grams lower compared
to single pregnancies. The decrease in birth weight halves from
triplet to twin pregnancies (1800 versus 900 grams). In the same
perspective, female births tended to yield lower birth weights up to
130 grams less compared to male babies.
Table 2: The effect of socio-economic determinants on birth weight.
Determinants of low birth weight
On the one hand, women’s education attainment, parity, and
wealth index were predictors of higher likelihood of normal birth
weight. With reference to poorest women, the likelihood that a
woman gives birth to a normal weight baby increases with wealth
quintiles to nearly double for women belonging to richer quantiles.
Similarly, higher education level predicted higher likelihood of
normal birth weight. For instance, women who attained tertiary
education were about twice more likely to deliver normal weight
babies compared to their counterparts who did not attend schooling.
Furthermore, compare to women with one parity, the likelihood
that a woman gives birth to a normal weight baby increases with
high parity to become two times more and nearly three times for
women with four to five and six and more parity respectively. On
the other hand, high blood pressure, multiple pregnancies, smoking
cigarette, and bearing a female child were negative predictors of
normal birth weight. In other words, chances of low birth weight
among women who reported smoking cigarette and those with high
blood pressure were more than twice likely compared to women
without the above conditions. Furthermore, the likelihood that
women give birth to underweight babies was nearly double among
women bearing female children or multiple pregnancies. Results of
the logistic model are summarized in Table 3.
Table 3: The effect of socio-economic determinants on low birth
weight.
Discussion
This study used secondary data from the national Demographic
and Health Survey 2017 to determine LBW prevalence and explore
socioeconomic and demographic factors that predict LBW among
Burundian women. LBW in Burundi was found to be close to other
low- and middle-income countries namely Ethiopia and Iran. In
Burundi, LBW is nearly 10% against 10% and 9.4% in Ethiopia
[14] and Iran [15]; respectively. Conversely, LBW was found to
be much higher in other countries such as India with one fifth of
live births weighing less than 2,5 kilograms [16]. In our study,
significant determinants of birth weights included wealth, parity,
chronic health condition, pregnancy type (single versus multiple),
woman’s age, and child’s sex. On the one hand, birth weight
consistently increases with higher wealth quintiles, higher parity
levels, and having diabetes. Compared to poorest women, birth
weight increases of 50 grams for women belonging to poorer
families. This increase doubles (100 grams) for women from
middle class and again doubles for richest women who gave birth
to babies weighing 200 grams more. Similarly, second and third
order newborns weighed 220 grams more compared to first order babies. The increase in birth weight was consistent with higher
birth orders and reached more than 300 grams for babies born to
multiparous women from sixth pregnancy going forward. Diabetes
was associated with an increase of more than 700 grams. On the
other hand, multiple pregnancies, older age, and female babies were
significant predictors of decreased birth weight. The most appealing
evidence concerns multiple pregnancies which cause birth weight
to decrease of about 1900 grams for triplet babies and about 920
grams for twin babies. Women older than 35 years gave birth to
babies lower of 70 grams weigh and female newborns weighed
130 grams less compared to their male counterparts. Results from
the linear model corroborate with those implemented using the
logistic regression. Wealth index and higher parity determined the
likelihood of giving normal weight babies while smoking, multiple
pregnancies, child’s sex, and a woman’s chronic condition were
significant predictors of the likelihood of LBW. Higher birth orders
were associated with more than twice likelihood of bearing a normweight
baby compared to first pregnancies.
In the same perspective, wealthier mothers and those highly
educated were nearly two times more likely to give birth to normal
weight babies. However, similar to the linear model results, twin
pregnancies and female babies were nearly twice more likely to
be underweight. Additionally, there was a double chance of giving
birth to underweight babies among women who smoke cigarettes
during the course of pregnancy were and those who had high blood
pressure. Determinants of birth weight in Burundi corroborate with
evidence from other settings. In Sri Lanka for instance, wealthier
women and higher educated women were more likely to give
birth to normal weight babies [17]. In this study, other significant
determinants of birth weight were mother’s age at childbirth,
newborn’s sex, and parity level. Contrary to the context of Burundi
where antenatal care did not predict birth weight, Indian women
who completed at least four recommended ANC visits were more
likely to deliver normal weight babies [17]. Similar evidence has
also been found in India [18]. Furthermore, similar to our findings,
a wealth evidence has established the correlation between women
with chronic health conditions such as diabetes with macrosomia
[19, 20]. Multiple pregnancies were significant predictors of LBW
in Burundi and in other similar settings. The example is the study
by Taywade et. Al [21] in India where women bearing multiple
pregnancies were 21 more likely to give birth to underweight babies
[21]. Additionally, male sex was found to be a protective predictor
of LBW in India and Ghana [18, 22]. Smoking was also a negative
predictor as women who reported having smoked cigarette during
pregnancy were at high risk of LBW in Ethiopia [23] and in India
[21].
Conclusion
This study which used a nationally representative sample
of Burundian women of reproductive age yielded evidence on
LBW prevalence and factors affecting it. The prevalence of LBW
was considerable when compared to many countries around the
world (10%). Among important predictors affecting birth weight
included a women’s age, wealth, education, parity, chronic health
condition, smoking, pregnancy type (single versus multiple) and
child’s sex. Markedly, LBW was highly associated with multiple
pregnancies, female babies, the history of smoking over the course
of the pregnancy, and high blood pressure. Despite considerable
efforts put in place by the government to improve maternal and
child health, LBW remains a public health concern. This study
suggests the need for the implementation of targeted innovative
interventions to tackle identified LBW risk factors among women
of reproductive age. For instance, socio-economic interventions
targeting poor household may help to improve economic status
of women with an aim to reduce LBW prevalence among poorer
women. Our study used secondary data analysis which has limit
in establishing causal inference between predictors and the
study outcome. This shortcoming could be addressed by cohort
studies of pregnant women which have the possibility to unpack
factors determining LBW with a much stronger evidence than
cross-sectional surveys. Therefore, this study provides important
evidence and a threshold for further research works in the field.
Read More About Lupine Publishers Journal of Pediatrics and Neonatology Please Click on Below Link:
https://lupinepublishers-pedaitrics.blogspot.com/
