The Effect of Educational Intervention on Improvement of Breastfeeding Self-Efficacy: A Systematic Review and Meta-Analysis

Background Self-efficacy is an important psychological and motivational factor in breastfeeding, and it is a valuable framework that predicts breastfeeding outcomes and demonstrates maternal confidence in breastfeeding. The meta-analysis evaluated the effectiveness of educational interventions on improving breastfeeding self-efficacy (BSE). Methods The English and Persian databases including Medline, Embase, Cochrane Database of Systematic Reviews (CDSR), PubMed, Web of Science, Scopus, CINAHL, Sid, IRANDOC, and Marg-Iran were systematically searched for studies published from January 2005 to December 2020. The quality of studies was evaluated using the Cochrane risk of bias tool and the heterogeneity by I2 statistic. The extracted data were analyzed using RevMan 5 statistical software and presented using random effects standardized mean difference (SMD). The funnel plot was used for evaluating publication bias. Results Results from 40 RCTs showed that educational intervention had a positive effect on the BSE compared with the usual/standard care (pooled SMD = 1.20; 95% CI = 0.75–1.64, p value <0.001). The subgroup analysis indicated that the educational intervention was based on theory, group class format, direct method education, during the first week of postpartum, doing during pregnancy, on primiparous women, and health center setting, and the Asian region has a more effect on BSE than the others. Conclusion Breastfeeding education is considered an influential factor in the improvement of BSE. It is recommended that breastfeeding education should be continued for several weeks after childbirth for gaining its benefit. The Asian region has a more effect on BSE than the others. Therefore, it is important to add the values in content of education in each country.


Introduction
Breastfeeding has the health benefit and promotes physical and mental health of the mothers and their infants. So that infants who are exclusively breastfed for six months experience less morbidity than those who are partially breastfed [1]. erefore, breastfeeding as a unique method of feeding and growth of infants in any situation and region of the world is recommended exclusively for the first 6 months of life [2]. Despite global strategy targets in low-and middleincome countries, only 37% of children younger than 6 months of age are exclusively breastfed that is below the WHO recommended rate of 50% [1]. It is even less in developed countries.
Self-efficacy is an important psychological and motivational factor in breastfeeding [3] and it is a valuable framework that predicts breastfeeding duration and demonstrates maternal confidence of breastfeeding [4,5]. Consistent with Bandura's cognitive-social theory, self-efficacy as a cognitive dynamics process evaluates people's beliefs and their ability to perform healthy behaviors and contributes to their preventive behavior [6]. Dennis maintains that four important sources including performance accomplishment, vicarious experience, verbal persuasion, and physiologic responses can impress women's levels of BSE [4]. Each of them may affect the mothers perceive about her breastfeeding, and it informs her BSE [4].
Developing and applying effective educational programs for improving breastfeeding self-efficacy (BSE) are important concerns for health professionals, and it can help mothers initiate and maintain breastfeeding for six months after birth [7,8]. In particular, breastfeeding educational programs have shown a positive effect on maternal breastfeeding behavior, awareness and attitude toward breastfeeding, continuity of breastfeeding, and BSE of mothers at six months after childbirth recently [5,9,10]. However, other studies have been reported conflicting results [11,12]. In some studies, the effectiveness of educational programs on BSE has been reviewed. However, in these studies, the limit number of English language studies has been reviewed or the quality of the studies was not determined or type of study was included observational studies, or only one dimension of education, such as theory or telephone based, was considered [5,[13][14][15].
ere is a wide variety in terms of the type and time of educational intervention. erefore, reviewing the difference in setting, time, and type of educational intervention on breastfeeding self-efficacy is one of the important steps to provide practical support for developing effective educational programs and policy making. e aim of this systematic review and meta-analysis was to determine the effect of educational programs on BSE.

Search Strategy and Data Sources.
e English and Persian electronic databases including Medline, Embase, Cochrane Database of Systematic Reviews (CDSR), PubMed, Web of Science, Scopus, CINAHL, Sid, IRANDOC, and Mag-Iran were systematically searched for studies published from 2005 to December 2020 using the following search strategies in accordance with the Mesh browser keywords and free-text words: (Feeding * OR Breast * OR Breastfeeding OR "Breast Milk" OR "Human Milk" OR "Breast Milk" OR Lactation OR "Milk Secretion" OR Colostrum OR "Exclusive Breastfeeding") AND (Pregnancy OR Gestation OR "Pregnant Woman") AND (Postpartum OR Puerperium) AND ("Self-Efficacy" OR "Self-confidence" OR "Self-concept") AND (Education * OR "Health Education" OR Instruction * OR Training) AND ("randomized controlled trial" OR "controlled clinical trial" OR randomized * OR randomly * OR trial * OR groups). Moreover, reference lists of the identified articles were screened. e PRISMA checklist (Figure 1) was used for reporting the search result [16]. We used a search strategy, which has been developed in Medline and adapted for other databases. We also used manual approaches, in particular, hand-searching and perusing reference lists of articles to find additional studies for systematic review. Various grey literature databases (such as Open Grey SIGLE, NTIS, Global Index Medicus (GIM), Google Scholar, World Cat, UW Libraries Search), theses, and dissertations, as well as conference proceedings, were assessed for collecting unpublished data. is study was not registered in the PROS-PRO database.
is meta-analysis was performed and reported in adherence with Preferred Reporting Items for Systematic Reviews and Meta-Analysis.

Inclusion Criteria.
e studies with the following criteria were included: (1) healthy mothers with healthy baby, (2) randomized controlled trials (RCTs) with at least two groups (control and intervention) that aimed to measure mother's BSE, (3) various forms of breastfeeding education (for example, didactic teaching session, face to face, indirect, individual, group, peer support, and workshop) combined with or without other interventions, (4) comparison groups were assigned to usual care and standard care, and (5) study published from Juan 2005 to December 2020 with restricted English and Persian. Due to the fact that the review of interventions in the last 15 years is critical for developing breastfeeding promotion strategy, the period for searching for articles in this review is limited to 2005-2020 years.

Exclusion Criteria.
e studies with the following criteria were excluded: (1) the mothers with chronic/systemic disease, parents with preterm baby with/without admitted at NICU ward, and complication in the breast, such as mastitis, (2) a quasi-experimental study, meta-analysis, and cross-sectional and observational studies, (3) noneducational intervention such as kangaroo mother care, motivational interview, psychoeducational counseling approach, and relaxation tone, and (4) randomized controlled trials (RCTs) that measured the mothers' knowledge or practice of breastfeeding, or breastfeeding initiation or duration.

Outcome.
In this study, the main outcome was breastfeeding self-efficacy. e subgroup analysis was done based on the type of education (telephone, theory based, and group/ individual), time of education (pregnancy/postpartum), participants (primiparous/multiparous), follow-up period, region, study quality, and setting (hospital/health centers).

Data Extraction.
e titles and abstracts of the eligible studies were independently extracted and screened by two authors (AM and SY), and duplicates were also removed. After providing the full texts of the studies, data extraction was done using a structured form included the name of the author, year of publication, location of study, type of intervention, the time of education, setting, participants, lengths and frequencies of sessions, the comparison group, sample size, measurement instrument, and results. Disagreements of the extractors were resolved by discussion or consultation with the third person.
2.6. Quality Assessment. Quality and risk of bias appraisal was conducted based on the guidelines of Cochrane Collaboration's tool for assessing the risk of bias in randomized trials [17]. 31 studies were considered having low risk of bias ( Figure 2  Obstetrics and Gynecology International education (telephone, theory based, and group/individual), time of education (pregnancy/postpartum), participants (primiparous/multiparous), setting (hospital/health center), follow-up period, region, and study quality. Publication bias was assessed using the funnel plot, and Egger's and Begg's tests ( Figure 3).

Results
e current study was updated until December 2020. A total of 1919 articles were extracted in the primary search based on the search strategy from the following databases: Scopus (129), ISI (515), PubMed (750), Mag-Iran (221), Cochrane (208), and other sources (96). en, we excluded 640 duplicate articles, 1189 in the title and abstract review, and finally, 47 articles after the full-text review. Four papers had the inclusion criteria, we decided to exclude them from this analysis because one was a conference abstract, two papers had non-English language, and one paper due to lack of reporting the mean of self-efficacy, and we did not receive usable data from the author. A total of 40 articles with 5743 subjects that met the criteria for inclusion were retrieved in this systematic review. All extracted articles were published in Persian or English. e quality assessment of articles was performed by two reviewers, independently. A flowchart of the extracted articles and selection procedure is shown in Figure 1.

Characteristics of Included Studies.
e characteristics of the 40 RCT studies included in the analysis are summarized in Figure 2. Of these, significant increases (p < 0.05) in BSE were reported in 26 of the included studies, which received breastfeeding education, and 14 studies reported no effect of education on BSE. Regarding the format of classes, 12 studies were based on group education that seven of them were based on theory. In 19 studies, educational tools such as booklets, pamphlets, video, and fillip cart were used for education. Moreover, women in 20 studies were educated with phone messages or telephone calling for follow-up. Based on the timing of education, the majority of interventions (31 studies) were done during pregnancy. e time of postpartum follow-up was varied at discharge to 24 weeks after childbirth. Regarding the region of the study, 18 studies were conducted in Iran and 2 studies were in Australia and Canada. One study was conducted in 9 countries including Croatia, Denmark, Hong Kong, Iraq, Japan, Spain, ailand, and Turkey. e five and four study was conducted in the USA and Brazil, respectively. Regarding the study quality, one study had a score of 4, eight studies had score 5, twentyone studies had score 6, four studies had score 7, and six studies had a score of 8. In all studies, self-efficacy was measured using Dennis breastfeeding self-efficacy (Table 1).

Main
Results. After quality assessment of studies, the results of 40 studies were included in the meta-analysis. e overall results demonstrated that educational intervention has a positive impact on BSE (pooled SMD � 1.20; 95% CI � 0.75-1.64, p value <0.001) (Figure 4). e high heterogeneity was seen among included studies (I 2 � 98%, p < 0.001) ( Figure 4). Egger's and Begg's tests were conducted to explore the publication bias in our meta-analysis ( Figure 3). e funnel plot in Figure 3 shows symmetrical funnel plots among studies. ere was no significant publication bias in this study based on Egger's and Begg's tests (p � 0.790 and p � 0.107, respectively).

Sensitive Analysis.
For sensitive analysis, we excluded 9 studies that have low quality. e sensitive analysis showed that there were no obvious changes after excluding the studies ( Figure 5). erefore, our results were reliable.

Discussion
e results from the pooled RCT data highlight the positive impact of educational intervention on the self-efficacy of breastfeeding compared with the usual/standard care. However, substantial heterogeneity was high across the included studies.
e subgroup analysis showed that the educational intervention was based on theory, group class format, direct method education, doing during pregnancy, on primiparous women, and health center setting, and the Asian region has a more effect on BSE than the others. e time of postpartum follow-up for evaluating the effect of educational intervention on BSE was a considerable point in the subgroup analysis. Accordingly, the effectiveness of education up to 6 weeks' follow-up period was significantly more than the other period with low heterogeneity among the included studies. Despite the high rate of heterogeneity, education in Asia was more effective than in other regions. Galipeau et al. conducted a meta-analysis of 9 randomized control trials and quasi-experimental studies to evaluate the effectiveness of all types of prenatal interventions either educational, support, or psychosocial on breastfeeding self-efficacy up to 4-6 weeks. e included studies were published from 2006 to 2016 in two English or French languages. ey reported that overall interventions  Obstetrics and Gynecology International 13 had a positive effect on breastfeeding self-efficacy compared with usual care. However, the quality of the included study was low, and the heterogeneity was high. According to the subgroup analysis, they reported that interventions based on theory and direct education methods were more effective than the others. Overall, our result was consistent with the results of Galilean's study in the term of the type and method of education. However, the Galipeau study was not conducting subgroup analysis based on the type of study, and in other subgroups, the sample size was small [56]. Also, in 2019, Ghasemi et al. [8] conducted a systematic review of 21 both randomized control trials and quasi-experimental studies to evaluate the effectiveness of the theory-based intervention (educational and noneducational) on self-efficacy of Iranian women. e included articles were conducted on the Iranian population that was published in the Persian and English languages from 2010 to 2019. ey were reported that the breastfeeding self-efficacy of mothers in the theory-based intervention group was more than the routine care group regardless of educational and noneducational intervention. Overall, our results were consistent with the results of Ghasemi's study in terms of theory-based intervention. However, Ghasemi's study was not doing meta-analysis and included studies were only in the Iranian population.
Also, Brockway et al. [13] in August 2016 performed a meta-analysis on 11 studies, both randomized control trials and quasi-experimental studies. eir study examined the effect of all interventions, whether education, support, counseling, or even screening and mechanical interventions on the BSE. In the event that these interventions are not of the same type, this issue has effect on their results. On the other hand, they reported high heterogeneity among included studies that indicated low quality of included studies. en, a subgroup analysis showed that only education but not support has a positive impact on the BSE. In contrast to our results, they reported only intervention during the postpartum period but not prenatal intervention, improved BSE. It seems that the difference in the number of included studies in the meta-analysis and different inclusion criteria may contribute to this difference. eir study used only results of 3 studies during the postpartum and 2 studies during the prenatal period that were a combination of support and education. While in our study, 29 studies during prenatal and 11 studies during the postpartum period only with the educational intervention were included. However, we observe that education in healthcare centers was more effective than in the hospital settings, while in their results, education only in a combination setting (hospital and community) affected the BSE. Similar to their study, theory-based education was more effective at improving BSE.
We observe that theory-based education in comparison with non-theory-based education can more improve mother's BSE. erefore, it is better to use theory-based educational intervention for educating these women.
ese results are consistent with the theory of self-efficacy, which states that modeling with practice is an effective way to increase self-efficacy [4]. Our results are in line with a previous meta-analysis, which is conducted by Chipojola et al. on 23 randomized controlled trial studies in 2019 [57]. is study indicated the overall effectiveness of educational programs based on theory (the theory of breastfeeding self-efficacy or theory of planned behavior) on the breastfeeding outcomes, and BSE. A high heterogeneity was reported among the included studies that indicated low-quality evidence. e subgroup analysis showed that mothers who received education based on the breastfeeding self-efficacy theory, providing in a hospital setting and developing countries, had a significantly higher score of BSE. Our finding is inconsistent with their results that indicate that education in healthcare centers is more effective than hospital setting. Also, in Guo et al.'s study, there was not seen any significant difference in the term of class format, time of education, and mode of class (face-to face or mobile education) on the BSE. It seems that the differences in inclusion criteria (only theorybased education) and number of included studies are caused by these differences. Another meta-analysis indicated a higher rate of exclusive breastfeeding among mothers who underwent interventions based on the theory of planned behavior [58]. Based on this theory, breastfeeding behaviors are influenced by attitudes, subjective norms, and perceived behavioral control [58]. During these sessions, mothers were taught the importance of breastfeeding and to create a positive attitude with invited influential people, as well as improved their perception of breastfeeding support [58].
One of the important findings of this study was that the effect of education on the BSE was significantly greater in Asian countries than in other countries. erefore, it is important to pay attention to breastfeeding values in each country. Factors such as religion, tradition, culture, beliefs, and customs can affect breastfeeding self-efficacy.
Moreover, we observe that the most effective education on BSE was in the first week after discharge. is relationship was seen even up to the 24th week of childbirth. erefore, it is important to continue education to increase women's BSE.

Strengths and Limitations.
One main strength of the current systematic review was that a large number of electronic databases, as well as hand-search, were comprehensively explored to yield maximum relevant articles on this field. However, the quality assessment of methodology and data extraction were done by multiple reviewers. Also, the subgroup analysis was performed based on several factors that previous studies did not consider them. Furthermore, we excluded quasi-experimental studies, and analysis was done only on RCTs; therefore, the best quality of evidence was available.
Despite the study strengths, the results have inherent limitations. In the first place, significant heterogeneity was seen among the included studies. Even after sensitive analysis, the level of heterogeneity was high; therefore, the quality of evidence is low. Although we attempted to make the included studies be like regarding the methodology (only RCTs) and type of intervention (only education), it is not fully achieved even under ideal conditions. Furthermore, regarding the inherent of this study, publication bias is inevitable due to some nonsignificant results that might not have been published. However, included articles that were published only in English and Persian may limit the generalizability to other populations, and this might cause selection bias.

Conclusion
Breastfeeding education is considered an influential factor in the improvement of BSE. It is recommended that these interventions are better based on the theory, in healthcare setting, a group class format, during pregnancy, with direct method format, and continued to the first week of postpartum. Considering these issues in designing, an educational intervention provides an important opportunity for health professionals to increase mothers' confidence for breastfeeding when they encounter breastfeeding problems.

Data Availability
e data used to support the findings of this study are available upon request.

Ethical Approval
Since it is systematic review and meta-analysis, no ethical approval was needed.