Prevalence and Determinants of Low Birth Weight in Ethiopia: A Systematic Review and Meta-Analysis

Department of Public Health, College of Medicine and Health Sciences, Wahemo University, Hossanna, Ethiopia College of Health Sciences, Debre Markos University, Debre Markos, Ethiopia School of Public Health, College of Medicine and Health Sciences, Bahir Dar University, P.O.Box 79, Bahir Dar, Ethiopia Adigrat University, College of Medicine and Health Science, Tigray, Ethiopia Department of Epidemiology and Biostatistics, Institute of Public Health, University of Gondar, Gondar, Ethiopia


Introduction
World Health Organization (WHO) defined low birth weight (LBW) as infants whose weight is less than 2500 gram at birth irrespective of gestational age [1]. It is one of the major public health problems globally, with a higher prevalence in developing countries [1][2][3][4]. Reports showed that 16% of all births globally were low birth weight, representing more than 20 million births a year [5,6]. Furthermore, LBW is a significant predictor of prenatal mortality and morbidity, especially in sub-Saharan African countries. It is also associated with a range of short-and long-term consequences including chronic disease in later life [1][2][3][4][5]7].
Several studies identified prematurity and intrauterine growth restriction as predominant causes of low birth weight. Prematurity is the leading reason in developed countries, whereas intrauterine growth restriction is in developing ones [1,6,8]. Additionally, inadequate weight gain during pregnancy, low prepregnancy weight, short stature, sex of the newborn (being female), hard physical work during pregnancy, illness(especially infections), the lower status of women, and lack of antenatal care (ANC) are also identified as risk factors for low birth weight in developing countries [9][10][11][12]. Multiple gestations, young maternal age, primiparity, low education level, poor maternal nutrition before and during pregnancy, and lower maternal anthropometric measurements are also associated with low birth weight [13][14][15][16][17][18][19][20].
Various studies were conducted to estimate the prevalence and determinants of low birth weight in Ethiopia. e prevalence of low birth weight in various studies conducted in Ethiopia ranged from 6% to 29.1% . Additionally, sex of the infant, pregnancy-induced hypertension, ANC follow-up, prematurity, parity, and residence were found to be associated with low birth weight .
However, most of these studies found inconclusive findings, especially on the factors associated with LBW. Moreover, most of the available studies in Ethiopia were conducted using small sample size limited to a certain region or area. erefore, this study was conducted to pool the available studies in the country and to present the national estimate of LBW and selected factors associated with it. e findings of this study will be important to design better health policies to prevent LBW and to design policies which can target the identified factors. Additionally, it will assist the country to provide information that can suggest areas of intervention and help achieve the national and international maternal and child health-related goals and targets.

Study Design and Search Strategy/Data Sources.
A systemic review and meta-analysis on the prevalence and determinant factors of low birth weight in Ethiopia were performed using the published and unpublished research articles. Major databases such as PubMed, Google Scholar, CINAHL, and African Journals Online (AJOL) were used to review published studies. In addition, we have used Google to retrieve unpublished studies from the official website of Addis Ababa University electronic database [51]. All published articles up to July 30, 2017 were included in this review. e reference list of already identified studies was also searched to retrieve additional articles. e online search was carried out using the following search terms and phrase: "prevalence of low birth weight, low birth weight, abnormal birth weight, newborn, neonate, underweight, infant, factors affecting low birth weight; pregnancy-induced hypertension (PIH) and low birth weight, parity and low birth weight, ANC and LBW, preterm delivery and LBW, and residence and LBW, Ethiopia" separately, and/or in combination using the Boolean operators to fit PubMed database, the following search strategy was applied: birth

Study Selection and Eligibility Criteria.
e following criteria were used to include and exclude studies from the review.

Inclusion Criteria
(i) Participants: this review included all studies, irrespective of characteristics of participants, that were conducted on prevalence and factors associated with low birth weight in Ethiopia (ii) Setting: all studies conducted in community or institution level (iii) Definition: all research studies that define LBW as birth weight less than 2,500 grams (iv) Publication types: journal articles, master's thesis, and dissertations (v) Studies published only in the English language were included in the review (vi) Timeframe: all studies irrespective of the time of data collection or publication were included

Exclusion Criteria
(i) Research studies that did not report outcome variable of this review were excluded (ii) Qualitative studies were excluded from the analysis (iii) Studies with methodological problems and review articles were also excluded from the review 2.3. Data Extraction. e data extraction was performed by five researchers using a data extraction tool. e researcher developed the data extraction tool. e tool included information on the name of the author, publication year, study design, data collection period, sample size, study area, the age of study participants, response rate, the definition used for low birth weight, the prevalence of low birth weight, and maternal and newborn-associated factors.

Quality Assessment and Data Collection.
Retrieved articles were assessed for inclusion using their title and abstract, and then, a full review of articles for quality assessment was performed before selecting for final review. Researchers used the Joanna Briggs Institute Meta-Analysis of Statistics Assessment and Review Instrument (JBI-MAStARI) for critical appraisal. Two reviewers independently assessed the articles for inclusion in the review using the instrument. e disagreement (on article selection and data extraction), which arose between the reviewers, was resolved through discussion and by involving a third reviewer. e quality of all studies included in this review after the final full article review was also assessed.

Publication Bias and Heterogeneity.
Publication bias and heterogeneity were also assessed. Egger's test was used to check the publication bias of studies. Statistically significant publication bias was declared at a p value less than 0.05. e heterogeneity of studies was tested using I 2 test statistics. I 2 test statistics result of 25%, 50%, and 75% was declared as low, moderate, and high heterogeneity, respectively. e random effect model was used for meta-analysis test results, which indicates statistically significant heterogeneity. Also, trim and fill analysis was conducted for studies with statistically significant publication bias using the random effect methods of analysis.

Statistical Methods and Analysis.
e analysis was conducted using STATA version 14 software. e forest plot was used to present the combined estimate and odds ratio (OR) with the 95% confidence interval. For studies, which did not present standard error (SE), it was calculated using the formula SE � �������� � px[1 − p]/n. e calculated standard error and prevalence rate of each study was then entered into STATA software to calculate the overall prevalence and its 95% CI. e pooled OR was used to show the effect of sex of neonates, pregnancy-induced hypertension, antenatal care booking, gestational age, parity, residence, and pregnancy intention on low birth weight.

Study Selection.
is systematic review and metaanalysis included all published and unpublished studies on LBW conducted in Ethiopia. Search for articles using databases resulted in a total of 1,004 articles. After screening all articles using their title, abstract, and full paper review, a total of 28 articles were selected for the final meta-analysis ( Figure 1).

Characteristics of Included Studies.
e mean birth weight of babies included in the review ranged from 2,077 to 3,147 grams, and the sample size ranges from 253 to 11,872. Overall, this review included a total of 50,110 newborn babies (Table 1). e majority [17] of the included studies were cross-sectional, 8 were cohort, and 3 were case-control studies. Most of the regions in Ethiopia (Oromia, Amhara, Tigray, SNNPR regions, Addis Ababa city, and Dire Dawa city) were represented in this review. Two of the studies were nation-based studies. e publication year of the studies was from 1990 to 2017 (Table 2). e quality of included studies was assessed using a quality assessment toot based on the JBI critical appraisal tool checklists for cross-sectional (19 studies), case-control (3 studies), and cohort (7 studies). Accordingly, the quality score of studies that used a cross-sectional study design ranged from 75% to 93.75%. For research studies that used cohort study design, the quality score ranged from 64% to 77%, and for case-control studies, the quality score ranged from 72% to 89% (Additional files 1-3).

Sex Difference in LBW.
Seven of the studies included in this review showed that the sex of the newborn is associated with low birth weight [24,25,30,39,[45][46][47]. In four studies, there was no significant difference in birth weight and sex of the newborn [26,34,43,48]. However, the final pooled estimate showed that female newborn babies are at a higher risk of low birth weight than males, OR � 1.46 (95% CI � 1.22, 1.73) ( Figure 3).

Preterm Delivery and LBW.
Six of the studies included in this meta-analysis showed that preterm babies (delivered before 37 weeks of gestation) were more likely to be low birth weight [29,34,40,45,46,50]. While one study showed the opposite finding [22], and one study showed no difference in prevalence of LBW between preterm and term babies [26]. is meta-analysis showed a significant association between preterm delivery and low birth weight, OR � 4.7 (95% CI � 1.53, 14.42) ( Figure 5).

Advances in Public Health
3.9. Residence and LBW. From 13 studies included in this category of meta-analysis [22,24-26, 29, 34, 40, 43, 45-49], two studies showed higher LBW risk among babies born from urban resident mothers [34,40]. But, nine studies showed an increased risk of LBW among rural residents [24-26, 29, 43, 46-49]. Two studies showed no statistically significant difference in the rate of LBW among rural and urban resident mothers [22,45]. e final pooled metaanalysis showed that babies born from rural mothers were at a higher risk of low birth weight compared to those born from an urban area, OR � 1.75 (95% CI � 1.19, 2.56) (Figure 8).

Heterogeneity and Publication Bias.
Egger's test was conducted to assess the publication bias of studies included in this meta-analysis. Accordingly, nonsignificant publication bias was observed in all categories of meta-analysis, p value >0.05. For studies that present with high heterogeneity between studies, the random effect method of analysis was used.

Meta-Regression.
e meta-analysis showed evidence of heterogeneity of included studies, and therefore, meta-regression was conducted to assess the source of heterogeneity. However, none of the covariates tested for source of heterogeneity were statistically significant. erefore, the heterogeneity could be explained by other variables not included in this meta-analysis (Table 3).

Discussion
is systematic review and meta-analysis were carried out to assess the pooled prevalence of low birth weight and its determinant factors in Ethiopia. e weight of the newborn at birth is considered as the most important factor related to perinatal and infant morbidity and mortality [52,53]. Low birth weight babies are almost 20 times more likely to die compared to normal birth weight babies [1].
is review found a higher prevalence of low birth weight in Ethiopia. e prevalence was 14.1% (95% CI � 11.2, 17.1). is is almost similar to the prevalence of LBW in sub-Saharan Africa, which ranges from 13% to 15% [1]. A study conducted in Cameroon showed a 13.5% prevalence of LBW [54]. A lower prevalence of LBW (8.7%) was also observed in a study undertaken in Colombia [53]. While a higher prevalence (30.3%) of LBW was found in a    e sociodemographic, economic, nutritional, and healthcare seeking behaviors of pregnant mothers and health service differences could be possible reasons for the higher and lower estimates.

Advances in Public Health
Ethiopia. Such disparities could be due to the difference in the sociodemographic, economic, healthcare seeking behavior, and nutritional status of pregnant women in different regions of the country. e subgroup analysis of the prevalence of LBW by publication year showed an increasing trend from 1990 to 2017. e rising prevalence in the LBW could be attributed to the increasing health care seeking behavior and reduced home delivery, which all increased the case detection rate over time in Ethiopia. e finding of this meta-analysis also showed the presence of a statistically significant association between the sex of the newborn and low birth weight, OR 1.5 (1.2, 1.7). Female newborn babies had a 46% higher odd of being low birth weight than male babies. is finding was congruent with the study conducted in Arunachal Pradesh [56]. e higher LBW risk among females than male newborn babies     Advances in Public Health could be due to the difference in chromosomal factors among male and female babies. Term male infants weigh between 150 and 200 g more than females [16]. Mothers who did not attend ANC for that pregnancy were 1.7 (OR 1.4, 2.2) times more likely to have low birth weight. Studies conducted in other countries also support this idea [16,[57][58][59][60]. Mothers who attended ANC are likely to get health education and counseling about pregnancy, which enables them to avoid modifiable risk factors for low birth weight. For example; studies have shown that prenatal interventions such as prevention of smoking during pregnancy, nutritional education, and prevention and early treatment of medical conditions during pregnancy are responsible for the reduction of low birth weight babies among women who follow antenatal care [60]. However, the effectiveness of prenatal care in reducing pregnancy complications depends on the number of visits and the time the women booked for prenatal care. erefore, studies showing the effect of frequency of antenatal follow-ups and the birth of the baby are needed.
In this meta-analysis, preterm babies were 4.7 (OR 1.5, 14.5) times more likely to be low birth weight compared to term babies. A baby born early has less time in the mother's uterus to grow and gain weight, and much of a fetus's weight is increased during the latter part of the mother's pregnancy [16].  Author (year) OR (95% CI) % weight

Advances in Public Health
Mothers who had pregnancy-induced hypertension were 6.7 times more likely to have low birth weight babies, OR 6.7 (3.5, 12.9). is finding is in line with the studies conducted in Karnataka, India, and Campinas, Brazil [61,62]. is might be due to the trophoblast invasion into the spiral arteries that supply the placenta is incomplete due to pregnancy-induced hypertension. e reduced uteroplacental blood perfusion leads to intrauterine growth retardation and low birth weight [63].
Parity is not associated with low birth weight in this metaanalysis. is could be because the studies included in this analysis have only two categories for parity (primipara and multipara). Previous studies showed that first and four and above births are more likely to be low birth weight compared to second and third births. e reason for this is that the uterus, especially the vascular structures, is not well matured for all purposes during the first pregnancy. e uterine vascular conditions start to decline increasing the risk of low birth weight beginning with the fourth pregnancy [16].
Maternal place of residence was associated with low birth weight (OR 1.6 (1.2-2.6)). Mothers living in rural areas had a 75% higher risk of giving low birth weight babies than mothers living in the urban areas. is finding is consistent with the study conducted in Punjab [64]. Rural resident pregnant women are more prone to maternal psychosocial stress, which is implicated in increasing the release of catecholamine and cortisol, and subsequently stimulates the release of corticotrophinreleasing hormone (CRH) through cortisol. e release of CRH has been hypothesized to initiate the onset of labor via a series of biochemical processes, while cortisol has been linked with intrauterine growth restriction and low birth weight [65].
Although this study has assessed the effect of selected variables on the occurrence of low birth weight, the factors are not limited to these. Previous studies have also reported additional factors associated with low birth weight. ese include poverty, nonuse of antenatal care, low social support and unwanted pregnancy [43], preterm birth, maternal chronic illness and maternal weight below 50 kg [29], and a history of perinatal death and  Author (year) OR (95% CI) % weight   [22]. erefore, future review and meta-analysis studies should consider addressing other factors not included in the current study.
is review used a comprehensive search strategy to retrieve both published and unpublished studies. Several authors were involved in each step of the meta-analysis. e review followed the PRISMA guideline for metaanalysis. On the other hand, because of the nature of the studies included, parity was dichotomized only into two, which may lead to a lack of association with low birth weight.
erefore, further studies are recommended to identify the effect of parity on low birth weight in Ethiopia. is study included the effect of selected factors on low birth weight. e factors are not limited to those included in this review, and further reviews to show the association of LBW with other sociodemographic, maternal, and newborn characteristics are needed.

Conclusions
e prevalence of low birth weight was high in Ethiopia, with high variation across the different regions in the country. is requires targeted interventions to regions with a higher prevalence of LBW. is review also showed the association of LBW with several factors. Variables such as sex of the newborn, prenatal care follow-up, pregnancy-induced hypertension, preterm delivery, and residence of the mother were significantly associated with LBW. No statistically significant association was observed in the rate of LBW among primigravida and multigravida mothers. Emphasis should be given to mothers living in rural areas. Community-based health promotion activities on maternal nutrition during pregnancy, antenatal follow-up, prevention, and early treatment of PIH are needed to reduce the high burden of LBW in Ethiopia.

ANC:
Antenatal care LBW: Low birth weight OR: Odds ratio PIH: Pregnancy-induced hypertension SNNPR: Southern Nations, Nationalities and Peoples' Region of Ethiopia WHO: World Health Organization.

Data Availability
e data used to support this study are included within this article.

Conflicts of Interest
e authors declare that they have no conflicts of interest

Authors' Contributions
NAK was involved in the design, selection of articles, study quality assessment, data extraction, statistical analysis, and writing the first draft of the manuscript. GMK, GAF, AAM, and AKB were also involved in the article selection, statistical analysis, and writing and editing of the manuscript. All authors approved the final version of the manuscript.