Determinants of Low Birth Weight Deliveries: an Unmatched Case-control Study in Five Referral Hospitals in Western Area Urban District, Sierra Leone

Background: Each year, about 20 million deliveries are Low Birth Weight (LBW) with 96.5% occurring in developing countries. The prevalence of LBW in Sierra Leone is about 7% while the Western Area Urban (WAU) district reports 17.5%. This study sought to determine factors associated with LBW deliveries in ve referral hospitals in the WAU district, Sierra Leone. Methods: A hospital-based unmatched 1:2 case-control study was conducted among mothers who delivered live singleton babies from November, 2019 to February, 2020 in ve referral facilities. Antenatal care cards of mothers were reviewed and a pre-tested questionnaire also administered to them. The association between maternal socio-demographic, socio-economic, obstetric and lifestyle factors and LBW was assessed using bivariable and multivariable logistic regression analyses. Results: A total of 438 mothers (146 cases and 292 controls), mean age: 24.2 (±5.8) and 26.1 (±5.5) years for cases and controls respectively participated in the study. Among mothers, 36.9 % (162/438) had no formal education, 25.8% (113/438) were unemployed and 29.7% (130/438) were unmarried. Multivariable analysis revealed that being unemployed (AoR = 2.52, 95% CI 1.16 - 5.49, p = 0.020), having anaemia during pregnancy (AoR = 3.88, 95% CI 1.90-7.90, p < 0.001), having less than two years inter-pregnancy interval (AoR = 2.53, 95% CI 1.11-5.73, p = 0.026), and smoking cigarettes during pregnancy (AoR = 4.36, 95% CI 1.94-9.80, p <0.001) were signicantly associated with having LBW babies. Conclusion: <


Background
Low birth weight (LBW) is of major public health concern especially in low and middle-income countries [1]. It contributes signi cantly to the burden of childhood diseases and increases mortality and disability in neonates and infants [2]. LBW is the weight of a newborn measuring less than 2.5 kg and is measured immediately after birth [3].
Newborns with LBW are at greater risk of neonatal mortality compared to normal weight babies [4]. Management of LBW deliveries increases health care costs due to long periods of hospitalization. It is estimated that care for extremely LBW babies is about six times more expensive compared to normal weight babies [5]. Evidence also exists that LBW increases the likelihood of developing noncommunicable diseases like diabetes and cardiovascular diseases in adulthood [5,6]. LBW has short term consequences, including respiratory distress, infections and hydrocephalus as well as long-term consequences, including coronary heart disease, diabetes, hypertension, and impaired cognitive function [7].
Globally, 15 to 20% of total births are LBW representing about 20 million births a year [8]. About 96.5% of LBW deliveries in 2019 occurred in developing countries; accounting for 60-80% of neonatal deaths [4].
The prevalence of LBW varies from region to region and within countries; with Asia accounting for 28%, Africa 13% and Latin America 9% [8].
In Sierra Leone, the pattern of LBW deliveries is unstable. The 2008 and 2013 Sierra Leone Demographic and Health Survey (SLDHS) showed that nationwide, LBW deliveries decreased from 11.0% in 2008 to 7.1% in 2013. However, it rather increased in the Western Area Urban district from 9.5-17.5% over the same period [9,10].
Several factors including early initiation of labour, mother's lifestyle (e.g. smoking and drinking alcohol), multiple pregnancies, maternal age < 15 years or > 35 years, poverty, infections and chronic illnesses including diabetes and hypertension have been reported to contribute to the delivery of LBW babies [6].
The purpose of the current study was to identify factors contributing to LBW in Sierra Leone to allow for the development of sustainable preventive measures in order to reduce the cost to the health sector. This study, therefore, assessed the socio-demographic, obstetric, maternal, and lifestyle factors associated with LBW babies delivered in ve referral hospitals in WAU, Sierra Leone.

Study design
A hospital-based unmatched case-control study was undertaken in ve referral hospitals in the Western Area Urban district of Sierra Leone from November, 2019 to February, 2020 among mothers who delivered singleton live babies. For each mother who delivered a LBW baby, two mothers who subsequently delivered normal weight babies in the same hospital were selected as controls.

Study setting
The study was conducted in ve-referral hospitals (Government and Non-governmental organization

Study variables
The independent variables included socio-demographic factors (age of mother, educational level, ethnicity, marital status and religion), socio-economic factors, (family income level and occupation), obstetric factors (maternal BMI, weight, height, parity, gravidity, abortion, ANC visits, gestational age, birth interval and anaemia during pregnancy), maternal health and lifestyle factors (diseases like hypertension, diabetes, heart disease, HIV, syphilis and malaria, alcohol intake, smoking and use of herbal medication).
The dependent variable was low birth weight.

Study participants and eligibility criteria
Mothers who delivered live singleton babies in the ve referral hospitals during the study period were eligible to participate in the study. The weight of every newborn was measured within an hour of birth.
Mothers with unknown last normal menstrual period, caesarean section birth, congenital deliveries, stillbirths, seriously ill, and mothers with singletons who weighed more than 4 kg were excluded from the study.
Cases de nition: Mothers who delivered babies weighing less than 2.5 kg and consented to take part in the study were enrolled as cases.
Control de nition: Mothers with newborns weighing 2.5 kg to 4 kg were enrolled as controls.

Sample size determination
An online OpenEpi software, version three for unmatched case-control studies was used to estimate the required sample size. A minimum detectable odds ratio of two and a percentage control group (birth spacing less than two years) exposed (67.6%) from a study conducted in Ethiopia was used [11]. A case to control ratio of 1:2 with a 95% con dence level and 80% power was used. The required sample size estimated was 438 (146 cases and 292 controls). sequentially as they delivered until the required sample size for the facility was attained. Each mother who gave birth to a singleton live baby weighing < 2.5 kg was enrolled as a case and two mothers who subsequently delivered singleton live babies weighing 2.5 kg to 4.0 kg were also enrolled as controls. In a situation where a case or control declined consent to be part of the study, the next eligible mother was selected as a replacement.
Data collection tools and procedure A questionnaire was developed speci cally for this study based on the objectives of the study and used for data collection from mothers who met the inclusion criteria of the study. The questionnaire was in three sections: socio-demographic and socio-economic, obstetric, and maternal health and lifestyle factors. The ANC cards of the mothers were reviewed and data on obstetric and maternal health factors extracted. Fifteen trained research assistants (three per hospital and one per shift) collected the data through face to face interviews using the common language used in the study area, 'Creole'. Midwives took the weights of newborn babies within one hour after delivery using calibrated Seca weighing scales. The research assistants administered the questionnaires to mothers within 24 hours post-delivery in a separate o ce within the ward to maintain privacy. Data were collected simultaneously in the ve hospitals until the required sample size was obtained.

Quality control
The questionnaire was developed speci cally for this study based on the objectives of the study, known variables from literature and the research experience of the authors. The questionnaire was pre-tested in three Government and NGO supported hospitals in the WAU district with similar service characteristics using 2% (three cases and six controls) of the sample size (cases). Findings of the pilot study were used to nalize the questionnaire. The data collected were checked daily for completeness and accuracy by a supervisor. The data were cleaned and entered in SPSS version 22 software and password protected.
Data processing and analysis Statistical analysis was done using Stata 15.0 (Stata Corp, College Station, TX USA). A histogram-normal curve was used to check the normality of continuous variables. Descriptive analysis i.e. summary statistics, mean (SD) and proportions were estimated, and inferential analysis using chi-square and logistic regression (bivariable and multivariable) to determine association between LBW and the independent variables. A stepwise backward elimination method with a restricted alpha level of 0.1(10%) was conducted and the post estimation command ("testparm i.variablename") used to determine variables that met the criteria for the multivariable logistic model. Odds ratio (OR) with 95% CI was computed and statistical signi cance was determined at p-value < 0.05.

Ethical consideration
Ethical clearance was obtained from the Sierra Leone Ethics and Scienti c Review Committee (SLESRC).
Permission was also obtained from the Chief Medical O cer, Ministry of Health and Sanitation (MoHS), and the hospital administration. The possible risk of the study (time) was explained to the mothers and their worries addressed. Written informed consent was obtained from each mother or her parent or guardian (mothers less than 18 years old) before data collection. The information collected was treated as con dential and codes were used to identify participants and not names.

Results
Socio-demographic characteristics of the mothers A total of 438 mothers (146 cases and 292 controls) were enrolled into the study from the ve-referral hospitals. They were aged 14-39 years, with mean age of 24.2 (SD: 5.8) years for cases and 26.1 (SD: 5.5) years for controls. One hundred and sixty-two of the mothers (36.9%) had no formal education, 25.8% (113) were unemployed and 29.7% (130) were not married (Table 1). Bivariable logistic regression analysis of the socio-demographic characteristics of the mothers revealed statistical differences between cases and controls in terms of age (CoR = 3.08, 95% CI 1.60-5.92, p = Obstetric characteristics of the mothers

Discussion
An unmatched case-control study was conducted among women who delivered in ve referral hospitals in the Western Area Urban district of Sierra Leone to determine factors associated with delivery of low birth weight (LBW) babies. This study has identi ed some socio-demographic, socio-economic, obstetric and lifestyle-related factors that are associated with the delivery of LBW babies in the study area.
Earlier studies from across Africa have reported some association between the delivery of LBW babies and several maternal socio-demographic, socio-economic, obstetric and medical factors [12][13][14]. In our current study, mothers who were unemployed were found to have a two-fold odds of giving birth to a LBW baby. This is consistent with ndings of a population-based study in southern rural Ghana, which established a strong association between birth weight and maternal employment status. Mothers who were unemployed and coming from the poorest households were reported to be more likely to give birth to LBW babies compared to those who were gainfully employed [12]. Unemployment can contribute to poverty, leading to poor maternal nutritional intake [14]. According to Ahmed, et al., maternal undernutrition and inadequate dietary diversity during pregnancy are signi cant determinants of delivery of low birth weight babies [15].
The odds of delivering a LBW baby as observed in the current study, was greater among mothers who were anaemic during pregnancy (Hb < 11.0 g/dl), compared to those who were not, similar to an earlier study also carried out in Ghana [7]. Other studies done in Ethiopia [4], Democratic Republic of Congo [14] and India [16] have also reported similar ndings in which anaemia during pregnancy was found to be associated with LBW. Anaemia during pregnancy can limit maternal oxygen uptake, thus reducing oxygen supply to the foetus and this contribute to foetal growth restriction [17].
In this study, babies born within less than two years after another child had higher odds of being of LBW, compared with those born after two or more years. Similar ndings have been reported from Ethiopia [18,19] and India [20]. This might be as a result of insu cient replacement of maternal nutrients used-up during the previous pregnancy and this may lead to reduced foetal development. A similar observation made in South Ethiopia by Mingude, et al. [13] has been explained on the basis that, the short interpregnancy interval is not able to allow the mother enough time to recover from the nutritional burden and stress of the previous pregnancy, resulting in maternal nutrition depletion. The short inter-pregnancy interval is also associated with maternal iron and folic acid depletion [15]. This reduces the ability of the mother to support foetal growth and development which in turn increases the possibility of growth restriction and LBW in subsequent pregnancies.
Our study also revealed that mothers who smoke cigarettes have higher odds of giving birth to a LBW baby. In a similar case-control study conducted in China [21], pregnant women who were exposed to even passive smoking had an increased risk of delivering low birth weight babies. Although, our study failed to determine the number of cigarette sticks smoked per day and for how long, other studies have reported that, mothers who are heavy smokers (> 8-10 cigarettes/day) had a higher odds of LBW babies [22].
Cigarette smoking is known to reduce oxygen supply to the foetus in-utero as carbon monoxide and nicotine-associated vasoconstriction reduces uterine and placental blood ow, thereby restricting the growth of the foetus, and hence can contribute to LBW [23]. Other factors including gravidity, maternal illhealth and taking of herbal medicine during pregnancy were not found to be signi cantly associated with the delivery of LBW babies in the current study after multivariable logistic regression analysis.
Although we employed a case-control study, which we consider an appropriate design for the study, the study is not without limitations. Assessment of some independent variables was liable to recall bias. The measurement we used for maternal smoking may not be the most suitable. Measuring the number of cigarettes sticks smoked by a mother in a day may better show the difference between the cases and controls. These limitations notwithstanding, the study has identi ed important maternal factors that when addressed, can reduce the incidence of LBW deliveries in Sierra Leone and other developing countries with similar characteristics.

Conclusions
This study has established that unemployment, anaemia during pregnancy, less than two years of interpregnancy interval, and cigarette smoking during pregnancy are signi cant factors for the delivery of LBW babies. There is a need for health care managers in WAU district to intensify screening and sensitization on the risk factors of LBW during antenatal sessions to address the identi ed factors among mothers. Also, the Ministry of Youth Affairs should design welfare programmes that would reduce the high rate of unemployment among women in Sierra Leone.