Objectives The aim of this systematic review and meta-analysis was to investigate the association of gender of newborn, antenatal care (ANC) and postnatal care (PNC) with timely initiation of breast feeding (TIBF) and exclusive breastfeeding (EBF) practices in Ethiopia.
Design Systematic review and meta-analysis.
Data sources To retrieve all available literature, PubMed, EMBASE, CINAHL, WHO Global Health Library, Web of Science and SCOPUS databases were systematically searched and complemented by manual searches. The search was done from August 2017 to September 2018.
Eligibility criteria All observational studies including cross-sectional, case-control, cohort studies conducted in Ethiopia from 2000 to 2018 were included. Newcastle-Ottawa Scale was used for quality assessment of included studies.
Data extraction and synthesis Study area, design, population, number of mothers (calculated sample size and participated in the study) and observed frequency data were extracted using Joanna Briggs Institute tool. To obtain the pooled effect size, a meta-analysis using weighted inverse variance random-effects model was performed. Cochran’s Q X2 test, τ2 and I2 statistics were used to test heterogeneity, estimate amount of total/residual heterogeneity and measure variability attributed to heterogeneity, respectively. Mixed-effects meta-regression analysis was done to identify possible sources of heterogeneity. Egger’s regression test at p value threshold ≤0.01 was used to examine publication bias. Furthermore, the trend of evidence over time was examined by performing a cumulative meta-analysis.
Results Of 523 articles retrieved, 17 studies (n=26 146 mothers) on TIBF and 24 studies (n=17 819 mothers) on EBF were included in the final analysis. ANC (OR=2.24, 95% CI 1.65 to 3.04, p<0.001, I2=90.9%), PNC (OR=1.86, 95% CI 1.41 to 2.47, p<0.001, I2=63.4%) and gender of newborn (OR=1.31, 95% CI 1.01 to 1.68, p=0.04, I2=81.7%) significantly associated with EBF. ANC (OR=1.70, 95% CI 1.10 to 2.65, p=0.02, I2=93.1%) was also significantly associated with TIBF but not with gender of newborn (OR=1.02, 95% CI 0.86 to 1.21, p=0.82, I2=66.2%).
Conclusions In line with our hypothesis, gender of newborn, ANC and PNC were significantly associated with EBF. Likewise, ANC was significantly associated with TIBF. Optimal care during pregnancy and after birth is important to ensure adequate breast feeding. This meta-analysis study provided up-to-date evidence on breastfeeding practices and its associated factors, which could be useful for breastfeeding improvement initiative in Ethiopia and cross-country and cross-cultural comparison.
Trial registration number CRD42017056768
- community child health
- primary care
- public health
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Strengths and limitations of this study
This systematic review and meta-analysis was conducted based on the registered and published protocol.
Since this is the first study in Ethiopia, the evidence could be helpful for future researchers, public health practitioners and healthcare policy-makers.
Almost all included studies were observational which might weaken the strength of evidence and hinder causality inference.
Perhaps, the results may not be nationally representative given that studies from some regions are lacking.
Based on the conventional method of heterogeneity test, a few analyses suffer from high between-study variation.
WHO and Unicef recommend timely initiation of breast feeding (TIBF) (ie, initiating breast feeding within 1 hour of birth) and exclusive breast feeding (EBF) (ie, feeding only human milk during the first 6 months)1 for maintaining maternal and newborn health.2 Breast feeding provides optimal nutrition, increase cognitive development, reduce morbidity and mortality for the newborn; for example, TIBF prevents 22% of neonatal deaths.3 Inappropriate breastfeeding practice, on the other hand, causes more than two-thirds of under-five child mortality, of which 41% of these deaths occur in Sub-Saharan Africa.1 4 Breast feeding also prevents maternal long-term chronic diseases, such as diabetes mellitus.3
According to a new 2017 global Unicef and WHO report, only 42% start breast feeding within an hour of birth, leaving an estimated 78 million newborns to wait over 1 hour to be put to the breast, the majority born in low-income and middle-income countries.5 The prevalence rate of TIBF varies widely across regions from 35% in the Middle East and North Africa to 65% in Eastern and Southern Africa. Another report also shows that only two in five infants <6 months of age are exclusively breast fed.6 The prevalence rate of EBF ranges from 22% in East Asia and Pacific to 56% in Eastern and Southern Africa.6 Based on our meta-analysis in 2018, the prevalence of TIBF and EBF in Ethiopia is 66.5% and 60.1% respectively.7 To date, globally, only 22 nations have achieved the WHO goal of 70% coverage in TIBF and 23 countries have achieved at least 60% coverage in EBF.2
To promote optimal breast feeding, WHO, Unicef and other (inter)national organisations have been working in developing countries, and several studies have been conducted on the advantages of breast feeding. However, it is still challenging to achieve the expected coverage and attributed to several factors including antenatal (ANC), postnatal care (PNC) and gender of newborn,8 9 and breastfeeding coverage continued to be suboptimal as a result. In Ethiopia, several meta-analyses studies were done on infant and young child feeding.7 10–14 In our previous meta-analysis, we explored the association between maternal employment, lactation counselling, mode of delivery, place of delivery, maternal age, newborn age and discarding colostrum breastfeeding practices (ie, TIBF and EBF).7 10 We also separately studied the association between TIBF and EBF.7 However, none of these meta-analyses did study the pooled effect of gender of newborn, ANC and PNC on TIBF and EBF. Given the absence of pooled estimates, up-to-date evidence is required to design intervention-based studies targeting these factors. Therefore, we aimed to investigate whether TIBF and EBF in Ethiopia are influenced by gender of newborn, ANC and PNC. We hypothesised at least one ANC or PNC visit significantly improves TIBF and EBF practices. Additionally, mothers with male newborn have higher odds of TIBF and EBF compared with mothers with female newborn.
Protocol registration and publication
The study protocol was registered with the University of York, Centre for Reviews and Dissemination, International prospective register of systematic reviews (PROSPERO) and published.15
Search strategy and databases
PubMed, EMBASE, CINAHL, WHO Global Health Library, Web of Science and SCOPUS electronic databases were searched to extract all available literature. The search strategy was developed using Population Exposure Controls and Outcome (PECO) searching guide in consultation with a medical information specialist (online supplementary file 1). The search was done from August 2017 to September 2018. Grey literature and cross-references of included articles and previous meta-analysis were also hand searched.
All mothers with newborn up to 23 months of age.
Gender of the newborn, ANC and PNC visit (at least one visit).
Female newborn, no ANC visit and no PNC visit.
TIBF and EBF practices.
Inclusion and exclusion criteria
Studies were included if they met the following criteria: (1) observational studies including cross-sectional, case-control, cohort studies; (2) conducted in Ethiopia; (3) published in English language and (4) published between 2000 and 2018. Studies were excluded on any one of the following conditions: (1) conducted in women with HIV/AIDS, preterm newborn and newborn in intensive care unit; (2) published in language other than English; (3) abstracts without full text and (4) qualitative studies, symposium/conference proceedings, essays, commentaries and case reports.
Selection and quality assessment
Initially, all identified articles were exported to Refwork citation manager (RefWorks 2.0; ProQuest LLC, Bethesda, Maryland, USA, http://www.refworks.com), and duplicate studies were cancelled. Next, a pair of independent reviewers identified articles by analysing the title and abstract for relevance and its compliance with the proposed review topic. Agreement between the two reviewers, as measured by Cohen’s Kappa,16 was 0.76. After removing irrelevant studies through a respective decision after discussion, full texts were systematically reviewed for further eligibility analysis. Newcastle-Ottawa Scale (NOS) was used to examine the quality of studies and for potential risk of bias.17 In line with the WHO standard definition, outcome measurements were TIBF (the percentage of newborn who breast feed within the first hour of birth) and EBF (the percentage of infants who exclusively breast fed up to 6 months since birth). Finally, Joanna Briggs Institute (JBI) tool18 was used to extract the following data: study area (region and place), method (design), population, number of mothers (calculated sample size and participated in the study) and observed data (ie, 2×2 table). Geographic regions were categorised based on the current Federal Democratic Republic of Ethiopia administrative structure.19 Disagreement between reviewers was solved through discussion and consensus.
A meta-analysis using a weighted inverse variance random-effects model was performed to obtain a pooled OR. In addition, a cumulative meta-analysis was done to illustrate the trend of evidence regarding the effect of gender of newborn, ANC and PNC on breastfeeding practices. Publication bias was assessed by visual inspection of a funnel plot and Egger’s regression test for funnel plot asymmetry using SE as a predictor in mixed-effects meta-regression model at a p value threshold ≤0.010.20 Duval and Tweedie trim-and-fill method21 was used to manage publication bias. Cochran’s Q X2 test, τ2 and I2 statistics were used to test heterogeneity, estimate amount of total/residual heterogeneity and measure variability attributed to heterogeneity, respectively.22 Mixed-effects meta-regression analysis was done to examine the effect of variation in study area (region), residence of women, sample size and publication year on between-study heterogeneity.23 The total amount of heterogeneity (R2)accounted for these factors was calculated by subtracting the residual amount of heterogeneity from the total amount of heterogeneity and dividing by the total amount of heterogeneity. Moreover, to assess the moderation effect of these factors, Omnibus test of moderators was applied. The data were analysed using ‘metafor’ packages in R software V.3.2.1 for Windows.23
Data synthesis and reporting
We analysed the data in two groups based on outcome measurements (ie, TIBF and EBF). Results are presented using forest plots. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline was strictly followed to report our results.24
Minor post hoc protocol changes
Based on the authors’ decision and reviewers' recommendation, the following changes were made to our published protocol methods.15 We added the JBI tool18 to extract the data. In addition, we used the Duval and Tweedie trim-and-fill method to manage publication bias. Furthermore, cumulative meta-analysis and mixed-effects meta-regression analysis were done to reveal the trends of evidence and identify possible sources of between-study heterogeneity, respectively.
Patient and public involvement
The research questions and outcome measures were developed by the authors (TDH and NTS) in consultation with public health professionals and previous studies. Given this is a systematic review and meta-analysis based on published data, patients/study participants were not directly involved in the design and analysis of this study. The results of this study will be disseminated to patients/study participants through health education on factors affecting breast feeding and disseminating the key findings using brochure in the local language.
In total, we obtained 533 articles from PubMed (n=169), EMBASE (n=24), Web of Science (n=200), SCOPUS (n=85) and CINHAL and WHO Global Health Library (n=5). Fifty additional articles were found through manual search. After removing duplicates and screening of titles and abstracts, 84 studies were selected for full-text review. Of these, 43 articles were excluded due to several reasons: 19 studies on complementary feeding, 3 studies on prelacteal feeding, 3 studies on malnutrition, 17 studies with different variables of interest and 1 project review report. As a result, 41 articles fulfilled the inclusion criteria and used in this meta-analysis: 17 studies investigated the association between TIBF and gender of newborn and ANC whereas 24 studies between EBF and gender of newborn, ANC and PNC. The PRISMA flow diagram of literature screening and selection process is shown in figure 1. One study could report more than one outcome measures or associated factors.
As presented in table 1, 17 studies reported the association of TIBF and gender of newborn and ANC in 26 146 mothers. Among these studies, 13 of them were conducted in Amhara (n=5), Oromia (n=4) and Southern Nations, Nationalities and Peoples’ (SNNP) (n=4) region. Regarding the residence status, eight studies were conducted in both urban and rural whereas six studies in urban women. All studies passed the NOS quality assessment criteria at a cut-off value ≥7.
Twenty-four studies reported the association between EBF and gender of newborn, ANC and PNC in 17 819 mothers. Of these studies, 11 were conducted in Amhara and seven in SNNP region. Based on the residence status, 10 studies were conducted in urban, 8 in urban and rural, and 6 in rural women. Even though almost all studies were cross-sectional, five studies have used nationally representative data of the Ethiopian Demographic Health Survey.19–23 Detailed characteristics of the included studies are shown in table 2.
Timely initiation of breast feeding
Among the 17 selected studies, 10 studies25–34 reported the association between TIBF and gender of newborn in 16 411 mothers (table 1A). The pooled OR of gender of newborn was 1.02 (95% CI 0.86 to 1.21, p=0.82, I2=66.2%) (figure 2). Mothers with male newborn had 2% higher chance of initiating breast feeding within 1 hour of birth compared with female newborn although not statistically significant. There was no significant publication bias (z=0.41, p=0.68) (online supplementary figure 1).
Likewise, 13 studies27 28 30 31 33–41 reported the association between TIBF and ANC in 12 535 mothers (table 1B). The pooled OR of ANC was 1.70 (95% CI 1.10 to 2.65, p=0.02, I2=93.1%) (figure 3). Mothers who had at least one ANC visit had 70% significantly higher chance of initiating breast feeding within 1 hour of birth compared with mothers who had no ANC visit. There was no significant publication bias (z=0.96, p=0.34) (online supplementary figure 2).
Exclusive breast feeding
Out of the 24 studies included, 11 studies25 26 42–50 reported the association between EBF and gender of newborn in 6527 mothers (table 2A). The pooled OR of newborn gender was 1.08 (95% CI 0.86 to 1.36, p=0.49, I2=71.7%) (figure 4). Since significant publication bias detected (z=−3.64, p<0.001), we did Duval and Tweedie trim-and-fill analysis and calculated a new effect size for gender of newborn (OR=1.31, 95% CI 1.01 to 1.68, p=0.04, I2=81.7%) after including imputed studies (ie, estimated number of missing studies=4) (online supplementary figure 3). Therefore, mothers with male newborn had 31% significantly higher chance of exclusive breast feeding during the first 6 months compared with mothers with female newborn.
Twenty-one studies35–37 42–49 51–60 reported the association between EBF and ANC in 16 052 mothers (table 2B). The pooled OR of ANC was 2.24 (95% CI 1.65 to 3.04, p<0.0001, I2=90.9%) (figure 5). Mothers who had at least one ANC visit had 2.24 times significantly higher chance of exclusively breast feed compared with mothers who had no ANC visit. There was no significant publication bias (z=1.69, p=0.09) (online supplementary figure 4).
Furthermore, seven studies42 44 49 53 54 56 60 reported the association between EBF and PNC in 2995 mothers (table 2C). The pooled OR of PNC was 1.86 (95% CI 1.41 to 2.47, p<0.0001, I2=63.4%) (figure 6). Mothers who had at least one PNC visit had 86% significantly higher chance of exclusively breast feed during the first 6 months compared with mothers who had no PNC follow-up. There was no significant publication bias (z=−0.91, p=0.36) (online supplementary figure 5).
As illustrated in figure 7, the effect of gender of newborn (figure 7) has not been changed whereas the effect of ANC on TIBF (figure 8) has been increasing over time.
Similarly, the effect of gender of newborn on EBF (figure 9) has not been changed over time. The effect of ANC (figure 10) and PNC (figure 11) have been increasing.
In studies reporting the association between TIBF and ANC, 26.29% of the heterogeneity was accounted for the variation in study area (region), residence of mothers, sample size and publication year. Based on the omnibus test of moderators, however, none of these factors influenced association between TIBF and ANC (QM=11.57, df=8, p=0.17). In studies reporting the association between TIBF and gender of newborn, the estimated amount of total heterogeneity was substantially low (tau2=4.28%); as a result, it is not relevant to investigate the possible reasons for heterogeneity.
Among studies reporting the association between EBF and gender of newborn, ANC and PNC, 77.66%, 60.29% and 100% of the heterogeneity were accounted for the variation in study area (region), residence of mothers, sample size and publication year, respectively. Based on the omnibus test of moderators, study area (region) and publication year negatively influenced the association between gender of newborn and EBF practice (QM=18.46, df=7, p=0.01). Study area (region) negatively influenced the association between ANC and EBF practice (QM=27.55, df=8, p=0.001) (table 3).
This meta-analysis assessed the association between breastfeeding practices (ie, TIBF and EBF) and gender of newborn, ANC and PNC. The key findings were EBF was significantly associated with ANC, PNC and gender of newborn whereas TIBF was significantly associated with ANC but not with gender of newborn.
In congruent with our hypothesis and the large body of global evidence,61–66 our finding indicated that mothers who had at least one antenatal visit had a significantly higher chance of initiating breast feeding within 1 hour of birth and exclusively breast feed for the first 6 months compared with mothers who had no ANC visit. This may be because health professionals provide breastfeeding guidance and counselling during ANC visit.7 The Ethiopian Ministry of Health has also adopted Baby-Friendly Hospital Initiative programme as part of the national nutrition programme and is now actively working to integrate to all public and private health facilities and improving breastfeeding practice as a result.
We also showed that mothers who had at least one PNC visit had nearly twice higher chance of exclusively breast feeding during the first 6 months compared with mothers who had no PNC follow-up. This result supported our hypothesis, and various studies have similarly reported a significantly high rate of EBF in mothers who had a postnatal visit at health institution66 or postnatal home visit.67 The possible justification could be that postnatal visit health education may positively influence the belief and decision of the mothers to exclusively breast feed. Previous studies have also shown that postnatal education and counselling are important to increase EBF practice.68 In addition, in our previous meta-analyses, we showed that guidance and counselling during PNC was significantly associated with high-rate EBF.7 Furthermore, PNC may ease breastfeeding difficulty, increase maternal confidence and encourage social/family support which lead the mother to continue EBF for 6 months.
Finally, in agreement with our hypothesis and previous studies,69–71 we uncovered gender of newborn was significantly associated with EBF practice. Mothers with male newborn had a 31% significantly higher chance of exclusively breast feeding during the first 6 months compared with mothers of female newborn. This finding disproved the traditional perception and belief in Ethiopia that male newborn has prelacteal feeding to be strong and healthy compared with female newborn. On the other hand, several studies63 66 depicted that gender of newborn is not significantly associated with breastfeeding practice, such as TIBF as we showed in our meta-analysis. This discrepancy might be due to the sociocultural difference and lack of adequate power given that we only found 10 studies to estimate the pooled effect size.
This systematic review and meta-analysis was conducted based on published protocol,15 and PRISMA guideline for literature reviews. In addition, publication bias was quantified using Egger’s regression statistical test and NOS was used to assess the quality of included studies. Since it is the first study in Ethiopia, the evidence could be helpful for future researchers, public health practitioners and healthcare policy-makers. The inclusion of all previously published studies is a further strength of this meta-analysis. This study has limitations as well. Almost all included studies were observational, which weakens the strength of evidence and hinder causality inference. Even though we have used broad search strategies, the possibility of missing relevant studies cannot be fully exempted and the finding may not be nationally representative. Based on the conventional method of heterogeneity test, a few analyses suffer from high between-study variation. The course of heterogeneity was carefully explored using meta-regression analysis, and this variation may be due to the difference in study area (region), residence of mothers, sample size, publication year or other residual factors; therefore, the result should be interpreted with caution. Moreover, the dose–response relationship between the number of ANC or PNC visits and breastfeeding practices was not examined. Lastly, significant publication bias was detected in studies that reported the association between EBF and gender of newborn. We did Duval and Tweedie trim-and-fill analysis to adjust publication bias and to provide an unbiased estimate; however, the result should be cautiously interpreted.
In line with our hypothesis, we found that increasing the use of antenatal and PNC has a positive effect on breastfeeding practices (ie, TIBF and EBF), which signifies stakeholders would provide emphasis on ANC and PNC service to optimise breast feeding. This meta-analysis study provided an overview of up-to-date evidence for public nutrition professionals and policy-makers in Ethiopia. It could also be useful for breastfeeding improvement initiative in Ethiopia and cross-country and cross-cultural comparison. From the research point of view, in general, intervention and outcome based studies on breast feeding in Ethiopia are required.
Our special gratitude forwarded to Sjoukje van der Werf (University of Groningen, the Netherlands) for her support to develop the search strings and Balewgizie Sileshi (University of Groningen, the Netherlands) for his support during the title and abstract screening.
Contributors NTS and TDH conceived and designed the study. TDH developed a syntax for searching databases, analysed the data and interpreted the results. TDH and SMA wrote and revised the manuscript. All authors read and approved the final manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement All data generated or analysed in this study are included in the article and its supplementary files.
Patient consent for publication Not required.