Determinants of Low Birth Weight Among Newborn Delivered At Public Hospital in Silte Zone, Southern Ethiopia: Case Control Study


 Background: There is high incidence of neonatal death in Ethiopia. There are various factors that contribute to neonatal death. Low birth weight (LBW) contributes for about half of infant death. Identification of modifiable determinants of LBW for potential interventions has received little attention in southern parts of Ethiopia. The aim of this study was to identify determinants of LBW among mothers delivered at hospitals in Silte Zone, Southern Ethiopia. Methods: A hospital-based unmatched case-control study was conducted in all public hospitals of Silte Zone. In the study, the number of mothers, cases and controls from each hospital was recruited using probability proportion to population size. Baby born with weight less than 2500gm were taken as a case where as a preceding three baby with birth weight of more than or equal to 2500gm was taken as control during study time in consecutive manner. Data was collected using pretested structured questionnaire, and by anthropometric measurements. Data was entered in to Epi Info 3.5.4 and exported to SPSS Version 20 software. Descriptive analysis and binary logistic regression model were used to calculate Odds ratio with 95% confidence intervals to estimate the association between the dependent and independent variables. Variables with a p-value less than 0.05 were considered as statistically significant. Result: The mean ± SD of birth weight of was 2174.47gm ±266.2for cases and 3370.48gm ±446.48for controls. After using multivariate logistic regression analysis, mothers who did not receive iron folate during pregnancy [AOR=2.84(1.15,7.03)], mothers who had anemia (hemoglobin <11mg/dl) during pregnancy [AOR=1.03(1.12, 6.24)], mothers who did not take additional meal [AOR =3.25(1.64,6.44)], mothers who did not receive nutritional counseling during the current pregnancy [AOR =6.93(4.80,11.76)], maternal under nutrition [AOR= 3.62(2.64, 6.47)], hypertensive disorder related to pregnancy [AOR= 2.76(1.34,5.71)] and inadequate Minimum Dietary Diversity Assessment(MDD-W) [AOR=6.65(2.31, 10.16)] were found as an independent and significant predictors of low birth weight. Conclusions: Maternal under nutrition, mothers not receiving iron folate during pregnancy, mother who had anemia, lack of target nutritional counseling, mothers who did not receive additional food during pregnancy, hypertensive disorder related to pregnancy and inadequate MDD-W were significant determinants of LBW. The importance of nutritional counseling with iron folate supplements, adherence during pregnancy and maternal under nutrition screening needs to be strengthened effort to reduce incidence of LBW infants.

LBW are characterized by high rates of morbidity and mortality due to infectious diseases, and are underweight, stunted or wasted in the start of the neonatal period and remain the same in their future life [2,3]. About 15-20% of all births worldwide are LBW, representing more than 20 million births a year.
More than 95% of these are born in developing countries. More than 50% of LBW infants are born in 13 developing countries that have the highest incidence available (20% or higher), whereas only 14% are born in 53 countries with an incidence of less than 10%,therefore, it is signi cant public health problem [4].
LBW is primarily due to babies born before 37 completed weeks of gestation (prematurity) and restricted fetal growth [2]. Studies indicate that preterm and intrauterine growth restrictions (IUGR) have strong relationship with LBW. The majority of LBW in developing countries including Ethiopia is due to IUGR which includes, poor maternal nutritional status at conception, chronic illness, chronic malaria, hypertension, inadequate dietary intake, short maternal stature due to the mother's childhood undernutrition and anemia whereas premature birth the most cause for LBW in developed countries [2,5].
In low socio-economic settings, various determinant factors play a great role on affecting birth weight and infants future health. LBW is highly affected by factors like maternal diet from birth to pregnancy, maternal nutritional status, maternal medical condition, pregnancy induced complication and sociodemographic factors contribute to poor fetal growth [1,3]. In the developing countries the magnitude of LBW could be underestimated due to high proportion of mothers give birth at home and newborns may not be registered as live birth and weighted (WHO, 2014). According to the 2005 Ethiopian demography and health survey (EDHS), 14% of babies were LBW and after ve years the prevalence decreased by 3% and it was 11% and the based on respondents report the proportion of very small sized babies in Southern Nation Nationality and People Region (SNNPR) was 15.2%In SNNPR only 29% of women delivered at a health care facility [6,7].
Multiple interventions involving monitoring progress of pregnancy, assessment of maternal nutritional status, iron and folic acid supplementation for pregnant mothers, maternal health and nutrition produce an average reduction in low birth weight incidence of 30% with greater effects on neonates [2][3][4][5].
Generally, preventive and control interventions for LBW morbidity and mortality should focus on: improved maternal nutritional status, prenatal care, micronutrient supplementation such as iron and folic acid, improved maternal health status and habits through treatment of complications, infections, health and nutrition education [2].
Hence, LBW is still a major public health problem-affecting infant, reducing their well-being and creating considerable demand for health service especially in developing countries. Therefore, the aim of this study was assess the determinant factors for LBW among newborns delivered at Silte zone, southern Ethiopia.

Study setting and Period
Silte zone is located 172km south of Addis Ababa the capital of Ethiopia and 202km from Hawassa capital of the south regional state. According to 2010E.C population projection, the zone consists of total population about 1,007,660 with male 498,792 (49.5%) and female 508,868 (50.5%) [7]. Based on health pro le of Silte zone health department, the zone has one comprehensive and specialized hospital, 3 primary hospitals, 34 health center with 195 health posts which are governmental and 20 private medium clinics, 45 private primary clinics, 01 pharmacy and 41 drug stores which deliver routine preventive and curative health services to the community. All public hospitals in the zone that include one comprehensive and specialized and three primary governmental hospitals were included in this study.

Study design and period
A hospital-based unmatched case-control study was conducted from March to April 30, 2018.

Source population
For cases: source populations were all neonates delivered or mothers who delivered with birth weight < 2500gm in public hospitals of Silte zone.
For the controls: source populations were all neonates delivered or mothers who delivered with birth weight ≥ 2500gm in public hospitals of Silte zone.

Study populations
For cases the study populations were mothers who delivered babies with birth weight of less than 2500gm at the study sites. For the controls the study populations were mothers who delivered babies with birth weight of ≥ 2500gm during study period at the study sites.

Inclusion criteria
For cases: babies who were born alive and birth weight of less than 2500gm were included in the current study.
For controls: babies who were born alive and birth weight of ≥ 2500gm were included in the study.

Exclusion criteria
For cases: neonates with birth weight of ≥ 2500gm, mother lack hemoglobin and Professional Initiated Testing and Counseling (PITC) data on card, mothers in severe health problems and diabetic mother.
For controls: neonates with birth weight less than 2500gm, mother lack hemoglobin and PITC data on card, mothers in severe health problems and diabetic mother.

Sample size determination
The sample size was determined using double proportion formula and calculated by Epi-Info version 3.5.4 statistical software package: with assumption of 95% con dence level 80% of power (Z β = 0.84), control to case ratio 3:1 (r = 3), the odds ratio to be detected 2 and the 24.6% control group was exposed. After adding 5% potential non-response rate the nal sample size was 404 (101 cases and 303 controls). Where: Note: Variables rural resident and birth interval 2 from Bale [8], MUAC taken from Kersa [9] and gestational age 37 weeks from Tigray [10].Sample size determination for predictors of LBW is presented in Table 1. Kibet and Alemgebeya Hospitals respectively) to total sample size of study. Then actual study participants were taken in respective consecutive single case, and preceding three normal birth babies as controls in each hospital until sample size gets its target during study period (Fig. 1).

Data Collection
The data was collected through interview using structured questionnaire adapted from related literatures [8, 9,11] and questionnaire was prepared in English and translated in to 'Amharic' language and back translated into English by independent persons to see consistency between the two versions by different language experts. It was used for collecting socio-demographic, maternal nutrition related factors, maternal health and reproductive health related factors. The questionnaire was pre-tested in Worabe town health center on 5% sample size before the actual data was collected to check for correctness and consistency of the questionnaire. Hemoglobin levels of each mother were taken from the card as hemoglobin and HIV test result was collected from mothers' card.

Anthropometric measurement
Standardization of anthropometric measurement was done by training the data collectors and comparing their measurement results with the results of the trainer, until the values of the data collectors and the trainer come to same measurement values. The weight of the newborns was measured within one hour upon delivery using a balanced Seca scale. The data collector measured mid-upper arm circumference (MUAC) of the mother two times and other data collector measured the same mother to see both intra and inter personal measuring differences, right after delivery using exible non-stretchable standard tape measure in cm. For left-handed women, the right arm was used instead. Height of mother was measured using digital machine. All measurements were taken twice and the average was taken.

Minimum Dietary Diversity Assessment (MDD-W)
MDD-W was collected using 24-hour recall method by MDD-W by use of the Global Dietary Diversity Indicator for Women [12].Brie y, the pregnant women were asked to recall the foods they had consumed in the previous 24 hour, probes to ascertain that no meal or snack was left out. A detailed list of all the ingredients of the dishes, snacks, or other foods consumed was generated to enable better classi cation of mixed dishes.

Operational De nitions
Birth weight: is the weight of the newborn measured after one hour of birth.
Intrauterine growth restriction (IUGR): describe a fetus that has not reached its growth potential because of maternal nutritional or maternal medical condition related factor.
Low birth weight: neonate with birth weight less than 2,500 grams.
Minimum Dietary Diversity-Women: proportion of women who received foods from ve or more food groups of the ten food groups [12] Multiple births: a mother gives two or more babies the same time in the current pregnancy.
Normal Birth Weight: newborn with a weight measurement of above or equal to 2500 grams Preterm birth: born alive before 37 completed weeks of pregnancy Singleton birth: a mother gives a single baby in current pregnancy.
Term baby: a baby delivered after 37 completed weeks and before 42 weeks.

Data quality control
Structured instruments that prepared in English, translated in to an 'Amharic' and back translated into English by independent persons to see consistency between the two versions by different language experts. The instrument was pre-tested in Worabe town health center on 5% of sample size before the actual data being collected. All data collectors and supervisors who are BSC holder was trained for two days by the principal investigator before the data collection on the objectives of the study and how to interview, measure, ll the questionnaire and how handle questions asked by participants. The principal investigator and 4-selected supervisors closely supervised the data collection process. Standardization of anthropometric measurement was done by training the data collectors and comparing their measurement results with the results of the trainer. Weighting scales was checked two times per day by measuring an object with a known and constant weight for quality control purposes. The collected data was checked, cleaned for completeness and consistency on daily basis by investigator and supervisors.

Data analysis and management
Data was checked for completeness and consistencies and then edited, coded and entered by using Epi-Info version 5.3.4 and exported to SPSS version 20 and checked for missing values before analysis.
Descriptive statistics was calculated for variables according to type. Frequency, means, medians and standard deviation was described for continuous variables after checking normality of data while the categorical variables was described by computing frequencies to determine the distribution of the variables in cases and controls. Bivariate and multivariable logistic regression analyses were computed to assess the association between study variables further to control all possible confounders. Then all variables signi cantly associated in bivariate analysis with p-value ≤ 0.25 or candidates was further entered into multivariable logistic regression and then multivariable analysis was performed to see association between dependent and independent variables and to adjust for all possible declared confounders. To estimate strength of association adjusted odds ratio (AOR) with 95% con dence interval was reported between study variables. A p-value less than 0.05 considered as statistically signi cant.

Results
Socio demographic data Data were collected from 404 mothers and their respective neonates with a response rate of 100%. The mean ± SD of birth weight was 2174.47gm ± 266.2for cases and 3370.48gm ± 446.48for controls. The majority of mothers among the cases (82.8%) and the controls (80.9%) were in the age group of 21-35 years followed by < = 20 years age of mothers 29.7% among the cases and 13.9% among controls.
Overall, most of the mothers in both cases (84.2%) and control groups (51.5%) were living in rural setting. More than half of (64.4%) mothers of LBW babies had no formal education and this accounted for 27.1% among the mothers of normal birth weight (NBW) babies (Table 2).

Determinants of LBW
In bivariate logistic regression analyses performed to identify candidate variables for multivariable logistic regression model in identifying determinants of low birth weight, all variables having p-value less than 0.25 were considered for the nal multivariable logistic regression model (Table 5). The selected independent covariates with p-value < 0.25 in bivariate logistic regression were entered into the multivariable logistic regression model to isolate the independent predictors of LBW using backward elimination stepwise likelihood ratio method. To control potential confounders such age, residence, educational status, maternal height, birth space, food taboo, coffee intake and malaria multivariate logistic regression was done statistical signi cant association was observed between nutritional counseling, iron and folate supplementation, anemia (hemoglobin < 11mg/dl), additional meals per day, women dietary diversity, maternal MUAC and hypertensive disorder of pregnancy. The result of the analysis showed that mothers who deliver female sex less likely 0.5 times to low birth baby than male (AOR = 0.44(0.11, 0.75)). Mothers who did not take iron and folate supplementation were more likely to 2.8 times deliver low birth weight babies than mothers who did take iron and folate supplementation during pregnancy (AOR = 2.84(1.2, 7.0)). In addition, anemic mothers (hemoglobin < 11mg/dl) were more likely to 1.03 times deliver low birth weight babies than mothers who had hemoglobin level of > = 11mg/dl during pregnancy (AOR = 1.03(1.12, 6.24)). Mothers who did not receive nutritional counseling during the current pregnancy were 6.9 times more likely to have low birth weight neonate (AOR = 6.93(4.80, 11.76)) than mother who did receive nutritional counseling. Similarly, mothers who did not take additional meal during the current pregnancy were more likely to have 3.3 times a neonate with low birth weight compared to those mothers who had additional meal during pregnancy (AOR = 3.25(1.64, 6.44)).
The odds of LBW among undernourished mothers 3.6 times higher as compared to their well-nourished counterparts was (AOR = 3.62(2.64, 6.47)). Also, mothers who had history of hypertension nearly three times odds of giving LBW neonates compared to non-hypertensive mothers (AOR = 2.76(1.34, 5.71)).
Furthermore, the odds of LBW babies among mothers with inadequate MDD-W were 6.65 times higher than mothers with adequate MDD-W (AOR = 6.65(2.31, 10.16)). Details of multivariate analysis were presented below (Table 6). Mothers who did not receive nutritional counseling during current pregnancy had 6.9 odds for giving LBW compared to mothers who did receive nutritional counseling. This might be due to the fact that nutritional counseling might improve their feeding behavior and hence, their nutritional status too which might in turn could help the mothers to decrease the risk of delivering LBW baby. This nding was consistent with other study conducted in Gondar University Hospital where mothers who didn't have ANC follow up had higher odds of giving LBW babies while this nding was attributed to the routine provisions of nutritional counseling during ANC visits [13].
The risk of low birth weight was 3.3 times among mothers who did not receive additional food during current pregnancy as compared to mothers who did receive additional food during current pregnancy. These ndings were consistent with other similar studies conducted in Jimma, Ethiopia and Ghana [14,15]. Mothers who consumed more food during pregnancy were 88% less likely to give birth to a low birth weight infant than those who ate the same as before pregnancy [15]. Mothers who did not take iron and folate supplementation were nearly 3 times at risk to deliver babies with LBW than mothers who did take iron and folate supplementation during pregnancy. This nding is in agreement with a study conducted in Bangladesh that showed intake of iron and folate supplements during pregnancy was found to have a protective effect against LBW [16]. These ndings were also consistent with other similar studies done in rural parts of Oromia, Ethiopia, Pakistan and a study from developed country [14,15,17]. Iron folate supplementation during pregnancy elevates hemoglobin level and prevents women from becoming anemic and consecutive increased risk of giving LBW babies because the required amounts may not be supplied from dietary intake during this period [16]. Furthermore, an overview of controlled trials suggested a 41% reduction in the prevalence of intrauterine growth retardation with folic acid supplementation, suggesting folate added to antenatal iron could independently affect birth weight of newborns [18].
Moreover, the current study showed that mothers with low level of hemoglobin were 1.03 times at risk to deliver LBW babies compared to mothers who had normal hemoglobin. This nding is consistent with other studies [16,19,20]. Mothers who did not receive iron folate and with low level of hemoglobin during pregnancy was more prone to give birth to LBW babies in Yemen [19]. It was also reported that maternal low hemoglobin level had a negative effect on birth weight among neonates in Bangladesh [16]. This might also be because of that micronutrient de ciencies during pregnancy had been shown to have serious implications on the developing fetus and hence, birth size [21]. Low level of hemoglobin could impair oxygen delivery to the fetus and thus interfere with normal intrauterine growth [22].
Inadequate women dietary diversity and maternal under nutrition during pregnancy independently and signi cantly affected LBW in our study. The consequences of inadequate nutritional intake and poor nutritional status for women during pregnancy not only directly affects women's health status, but may also have a negative impact on birth weight and early development. Mothers having inadequate MDD-W had 6.65 times odds of giving birth to LBW babies. This nding was consistent with a study done in rural Oromia region in Ethiopia, in which women in the inadequate MDD-W group had an increased risk of LBW and preterm baby compared with women in the adequate MDD-W group [14]. Similarly, a study from Ghana shows that Women dietary diversity score and dietary patterns were found to be protective against low birth weight [15]. However, a recent randomized controlled trial in India reported that an intervention that increased consumption of dairy, fruits, and green leafy vegetables before and during pregnancy through a specially formulated snack had no effect on birth weight [23]. This discrepancy might be due to differences in study population, geographical location, and the study design.
Similarly, maternal under nutrition as measured by maternal MUAC of less than 23cm signi cantly increased 3.6 times the odds of having LBW babies in the present study. This nding is in agreement with an observational cohort study conducted on 1295 live birth and pregnant women to measure the incidence and determinants of LBW in a rural population in Kersa district, Western Ethiopia, where maternal MUAC of less than 23cm was found to signi cantly predict LBW in this cohort [9]. Moreover, recent study from Yemen showed that maternal under-nutrition de ned by MUAC of less than 23cm was signi cantly correlated to LBW [19]. The observed ndings might be because of the fact that anthropometric measurements directly or indirectly measures nutritional status. Even though either acute or chronic maternal malnutrition has direct effect on the birth weight of a baby, acute maternal malnutrition has more pronounced effect [9].
Mother who had hypertensive disorder related to pregnancy was 2.8 times at risk for low birth weight in the present study. This nding is consistence with a case control study conducted in Malaysia showed that hypertensive mother had signi cant association with LBW infants [24]. This study agreement with another case control study done in India also showed that women with pregnancy-induced hypertension have 2.5 times more risk of giving birth to LBW babies [25].
limitation of the study There might be chances of recall bias as respondents had to remember past history like dietary diversity, food frequency, number of ANC visit, number of iron folate tablet taken.

Conclusion
Lack of nutritional counseling during pregnancy at ANC visit, lack of iron folate supplementation, female sex, low level of hemoglobin, lack of additional food intake during pregnancy, MUAC less than 23cm, hypertension during pregnancy and inadequate MDD-W were identi ed as signi cant predictors of LBW among neonates under the study. Availability of data and materials All necessary data are available within the paper.
Ethical approval and consent to participant The study was approved by the Institutional Review Board of Wolaita Sodo University, College of health sciences and medicine. Permission letter was obtained from the all study site. The objectives, expected outcomes, bene ts, and risks of the study were explained for study patricians. Data was collected after written informed consent was obtained. The study was conducted in accordance with the Declaration of Helsinki.
Authors' contributions MU Conceived and designed the study, execution of the study, data acquisition data analysis and interpretation, article preparation. YAW&MMA Data analysis, interpretation and prepared initial draft of the manuscript for publication. The nal manuscript was read and approved by all authors. All authors reviewed and agreed on the nal version of the article before submission.
25. Ghimire R., P.D.B., Phalke V.D., Banjade B., Singh AK. , Determinants of low birth weight: Acase control Study in Pravara Rural Hospital in Western Maharashtra, India. International Journal of Scienti c Research, 2014. 3(7). Figure 1 Schematic presentation of the sampling procedure in the study area