The impact of Baby Friendly Initiative accreditation: An overview of systematic reviews

Abstract Despite its reported benefits, breastfeeding rates are low globally, and support systems such as the Baby Friendly Initiative (BFI) have been established to support healthy infant feeding practices and infant bonding. Increasingly reviews are being undertaken to assess the overall impact of BFI accreditation. A systematic synthesis of current reviews has therefore been carried out to examine the state of literature on the effects of BFI accreditation. A systematic search of CINAHL, MEDLINE, Maternal and Infant Health, Scopus, the Cochrane Library and PROSPERO was undertaken. Study selection, data extraction and critical appraisal of included reviews using the AMSTAR‐2 tool were undertaken by two authors, with disagreements resolved through discussion with the third author. Due to heterogeneity, a narrative synthesis of findings was applied. Fourteen reviews met the inclusion criteria. Overall confidence in the results of the review was rated as high for three reviews, low for two reviews and critically low for nine reviews. Most evidence suggests some increase in breastfeeding initiation, exclusivity and duration of breastfeeding, and one main trial suggests decreased gastrointestinal infection and allergic dermatitis in infants. However, overall certainty in the evidence was rated as very low across all outcomes due to concerns over risk of bias within and heterogeneity between the original studies. More contemporary, good‐quality randomised controlled trials or well‐controlled prospective comparative cohorts are required to better evaluate the impact of full BFI accreditation, with particular attention paid to the context of the research and to long‐term maternal and infant health outcomes.

there is no evidence to move away from the guidance to exclusively breastfeed infants for the first 6 months of life (Smith & Becker, 2016).
Evidence suggests that infants fed with breastmilk substitutes are at increased risk of gastrointestinal infections, respiratory infections, asthma, coeliac disease and sudden infant death as well as increased risk of obesity and diabetes in later life (Lessen & Kavanagh, 2015;Victora et al., 2016). Numerous adverse outcomes are also increased in mothers who do not breastfeed their infants, including ovarian cancer, breast cancer, type 2 diabetes and postnatal depression (Gunderson et al., 2018;Lessen & Kavanagh, 2015;Victora et al., 2016). Worldwide, an estimated 823,000 deaths in children under five and up to 20,000 deaths a year from breast cancer could be prevented by improving breastfeeding practices . Although breastfeeding can protect against child and maternal deaths in low-, middle-and high-income countries, disparities in the magnitude of different health benefits according to country income level are known to exist .
Global data show the prevalence of breastfeeding at 6 and 12 months decreases with increasing national wealth, with prevalence of breastfeeding at 12 months decreasing 10% for each doubling in gross domestic product per head . As well as national wealth, breastfeeding rates are lower in women who are younger, of low socio-economic status, living in deprived areas, of lower educational attainment, who smoke (Cohen et al., 2018;McAndrew et al., 2012) and in women with a raised body mass index (Wojcicki, 2011).
Despite its reported benefits, breastfeeding rates up to 6 months or longer appear to be low globally , and systems such as the UNCIEF Baby Friendly Initiative have been established to support healthy infant feeding practices. The Baby Friendly Hospital Initiative (BFHI) was developed in 1991 and updated in 2018, with the aim for every baby to have the best start in life through the global protection, promotion and support of breastfeeding in facilities providing maternal and newborn services (United Nations Children's Fund & World Health Organization, 2018). Each facility is required to comply with the 10 steps to successful breastfeeding, which incorporate adherence to the WHO Code for Marketing of Breastmilk Substitutes, policy development, staff training and key clinical practices for supporting breastfeeding (see Appendix A for full details of these steps). The BFHI initiative has also been expanded to include a 7-point plan for community services to support sustained breastfeeding (UNICEF, 2014). To become accredited, community facilities are required to have a written breastfeeding policy, training for staff, and provide a supportive and welcoming atmosphere for breastfeeding women and work collaboratively with the aim for increased exclusivity and duration of breastfeeding (see Appendix B).
Each country adopts the Baby Friendly Initiative into its own framework for accreditation (for example BFHI Australia, 2020; UNICEF UK, 2017). Of the 155 countries included within a WHO survey, 71% had an operational BFHI programme in 2016-2017; however, only six countries reported that the majority of their facilities had BFHI accreditation (WHO, 2017). Overall, coverage was estimated to be 10%, although there were wide variations between and within regions-for example, in Europe over half of births occurred within BFHI accredited facilities within 13 countries, but 12 different European countries had no accredited facilities (WHO, 2017).
Many systematic reviews have evaluated evidence behind individual steps of the Baby Friendly Initiative (Jaafar, Ho, Jahanfar, & Angolkar, 2016;Jaafar, Ho, & Lee, 2016;Lumbiganon et al., 2016;Moore et al., 2016;Smith & Becker, 2016). Increasingly reviews have also been focusing on the assessment of the overall impact of BFI accreditation. As systematic reviews are increasingly published, clinicians can be left feeling overwhelmed by the plethora of evidence. Therefore, the requirement for overviews of reviews is gaining recognition to enable systematic reviews to be compared and the evidence collated to provide an overall understanding of the available information on a given topic (Aromataris et al., 2014). The aim of this overview was therefore to evaluate the quality and extent of systematic evidence regarding the impact of Baby Friendly Initiative accreditation in order to better understand the effectiveness of this global intervention on breastfeeding rates and health related outcomes. Consideration was given to the income level of the country of the original trials and the level of BFI accreditation when evaluating the evidence.

| Search strategy
Library and PROSPERO registry of systematic reviews was undertaken. The WHO website was also searched for relevant publications.
The search strategy included terms around 'baby friendly initiative' and 'systematic review'. An example full search strategy within one database can be found in Appendix C. Databases were searched from 1991 when the Baby Friendly Initiative was launched to 6 March 2020. Systematic reviews were limited to those published in the English language. Reference lists of included systematic reviews and other relevant literature were screened manually for further citations.

| Inclusion and exclusion criteria
Retrieved citations were screened by two independent researchers against the eligibility criteria (Table 1) by title and abstract and then full text for relevant articles; any disagreements were resolved through discussion with the third author. Authors of citations of conference proceedings or protocol registrations were contacted to enquire after full text articles.

| Data extraction
Data extraction was undertaken by two researchers using a predefined data extraction table. Authors of the systematic reviews were contacted where required for additional information.

| Risk of bias assessment
Two reviewers independently assessed included systematic reviews for risk of bias using the Assessment of Multiple Systematic Reviews v2 (AMSTAR-2) checklist (Shea et al., 2017). See Appendix D for the full checklist. Disagreements were resolved through discussion with the third reviewer. The checklist authors (Shea et al., 2017) believe seven domains within the checklist can critically affect the validity of the review and its conclusions but acknowledge that reviewers can add or substitute domains as required according to the nature of the systematic reviews appraised. Within this overview, nine AMSTAR-2 domains were considered to be critical. Seven of these coincide with areas considered critical by the checklist authors (Shea et al., 2017), including Item 2: review methods established prior to conducting review; Item 4: comprehensive literature search; Item 7: justification for excluding individual studies; Item 9: satisfactory techniques for assessing risk of bias within included trials; Item 11: appropriate methods for statistically combining results; Item 13: risk of bias considered when interpreting/discussing review results and Item 15: assessment for presence of publication bias. Given disparities in breastfeeding outcomes between countries and the wide range of sociocultural determinants that can impact upon breastfeeding Item 14: heterogeneity of included studies discussed was also considered critical for this review. Item 12: the impact of risk of bias considered on meta-analysis results was also considered a critical domain within this overview. An overall rating of confidence in the results was given depending on the presence of flaws in the above critical domains or other weaknesses identified within the systematic review in accordance with the criteria set out in Shea et al. (2017); high overall confidence where there was no or one weakness within a non-critical domain, moderate overall confidence where there was more than one weakness in a non-critical domain, low overall confidence where there was one critical weakness, with or without other weaknesses in non-critical domains or critically low confidence in the results where there was more than one weakness in a critical domain. For the purposes of this review, no weakness was considered to have occurred within the domain if the criteria were fully or partially met.

| Data synthesis
Using a narrative synthesis approach, a formal discussion of the results of the systematic review evidence base regarding the impact of Baby Friendly Initiative accreditation was undertaken. This

| Characteristics of included systematic reviews
Characteristics of the included systematic reviews can be found in Table 2. Thirteen systematic reviews looked at BFI within a hospital and/or community setting, with one examining breastfeeding promotion within the neonatal intensive care unit (NICU) setting . The included systematic reviews predominantly used a narrative synthesis approach as meta-analyses was inappropriate due to the high degree of heterogeneity within included studies in relation to study design, intervention and definitions of outcomes (initiation, duration and exclusivity). Only four systematic reviews performed meta-analyses (Chung et al., 2008;Kim et al., 2018;Sinha et al., , 2017. BFI was the sole intervention in four systematic reviews (Atchan et al., 2013;Fallon et al., 2019;Munn et al., 2016;Pérez-Escamilla et al., 2016) with the remainder exploring any interventions aimed at improving breastfeeding initiation, duration and exclusivity but included BFI as a subgroup. Two reviews exclusively included randomised controlled trial evidence or quasi-randomised studies (Chung et al., 2008;Kim et al., 2018), one review included quantitative studies and systematic reviews (Feltner et al., 2018), three included research of any methodology (Beake et al., 2012;Fallon et al., 2019;Munn et al., 2016) and one review included quantitative, mixed methods or systematic review studies (Hannula et al., 2008), with the remaining reviews incorporating quantitative research that included randomised or observational study designs.
The reviews included a total of 105 individual studies that were attributed by the reviews to be evaluating BFI implementation. The number of BFI implementation studies in each review ranged from 2 (Chung et al., 2008;Fairbank et al., 2000) to 58 (Pérez-Escamilla et al., 2016 Table 3 provides AMSTAR-2 quality assessment results for each included systematic review.

| Methodological quality of included systematic reviews
Only one review (Atchan et al., 2013) was judged not to have a clearly focussed research question and inclusion criteria.
One review (Feltner et al., 2018) reported a registered review protocol established prior to undertaking the review, and a further three reviews mentioned at least some aspects of a protocol (Fairbank et al., 2000;Fallon et al., 2019;Sinha et al., 2017).
The remaining reviews made no explicit reference to a review protocol.
The search strategy was judged to be inadequate in three reviews (Hannula et al., 2008;Kim et al., 2018;Munn et al., 2016), due to no clear justification for the date restrictions applied within their search.
Four reviews (Atchan et al., 2013;Hannula et al., 2008;Howe-Heyman & Lutenbacher, 2016;Munn et al., 2016) did not adequately report assessing risk of bias within included studies, and the assessment of risk of bias in non-randomised studies was deemed to have only partially used a satisfactory technique within three reviews (Fairbank et al., 2000;Sinha et al., , 2017. Of the four reviews that included meta-analysis (Chung et al., 2008;Kim et al., 2018;Sinha et al., , 2017, all justified the use of metaanalysis, used appropriate methods and explored the causes of heterogeneity within the results. However, none reported carrying out adequate investigations to determine the impact of potential publication bias on the results and one (Kim et al., 2018) did not report assessing the impact of risk of bias within the included studies on the meta-analysis results, for example, through sensitivity analysis.
One review (Hannula et al., 2008) did not discuss the likely impact of risk of bias within included studies when interpreting or discussing the review's results. All reviews were judged to have provided at least some exploration, explanation or discussion around heterogeneity of studies included within their review.
All but two reviews (Atchan et al., 2013;Hannula et al., 2008) declared any conflicts of interest for the systematic review; however, none of the studies reported on the funding sources of the included studies within the review.
Overall confidence in the results of the review was rated as high for three reviews (Fairbank et al., 2000;Fallon et al., 2019;Feltner et al., 2018) that only had one or no weaknesses in non-critical domains, low for two reviews (Beake et al., 2012;Renfrew et al., 2010) that had a weakness within one critical domain and critically low for the remaining nine reviews that had weaknesses within more than one critical domain. No reviews were rated as moderate.

| Quality of the studies included within the reviews
As well as assessing the quality of the included reviews, it was important to consider the quality of the studies included within the reviews as indicated in the Risk of Bias assessments by the review authors  F I G U R E 1 Risk of bias and study design as assessed by review author(s) of the 105 individual studies included within the reviews. † One prospective cohort was noncomparative included in multiple reviews, the original study was obtained for clarification. The methodology of some studies however remained ambiguous.
The most cited study was a randomised controlled trial (RCT) conducted in Belarus, appearing in 11 of the 14 reviews (Kramer et al., 2001). This was rated to have medium/low risk of bias.

| Outcomes
There was considerable overlap of the studies within the systematic reviews; therefore, where a study was included within multiple reviews, the results were only reported once in order to avoid duplication. Appendix G provides full results from each review.

| Breastfeeding intention
No review reported breastfeeding intention as an outcome.

| Breastfeeding initiation
Breastfeeding initiation was reported in 11 systematic reviews. The results of four reviews (Chung et al., 2008;Fairbank et al., 2000;Fallon et al., 2019;Hannula et al., 2008) have not been reported narratively to avoid duplication as all of the study(ies) contributing to this outcome within those reviews were already reported within another included review (Beake et al., 2012, Pérez-Escamilla et al., 2016. The other seven reviews reported outcomes from 32 studies; 17 of which were only reported within one review. No review reported this outcome from any RCT evidence. The majority of the included observational study evidence suggested breastfeeding initiation increased with BFI accreditation; however, this was not consistent across the included studies within all reviews, with some studies showing no difference in breastfeeding initiation with BFI accreditation (Beake et al., 2012;Feltner et al., 2018;Howe-Heyman & Lutenbacher, 2016).
This outcome was difficult to interpret however due to different definitions of exclusive breastfeeding within each study and review.
Evidence from three RCTs and some low-quality observational studies suggested good implementation of Step 10 (continued support after discharge, e.g., home peer support) was necessary for improvements in both exclusive or any breastfeeding (Pérez-Escamilla et al., 2016).

| Any breastfeeding
Thirteen reviews reported duration or rates of any breastfeeding. All of the included studies within five reviews (Chung et al., 2008;Fairbank et al., 2000;Hannula et al., 2008;Renfrew et al., 2010;Sinha et al., 2017) were either reported in another review (Pérez-Escamilla et al., 2016 or were not studies evaluating full The rate of continued breastfeeding from 6 to 23 months was no different between intervention and control groups . There was some evidence that the more BFHI steps that were implemented, the longer the duration of any breastfeeding (Atchan et al., 2013;Feltner et al., 2018;Pérez-Escamilla et al., 2016), with women who receive no BFHI practices 13 times more likely to discontinue breastfeeding prior to 6 months (Atchan et al., 2013). There was some evidence that implementation of Steps 2 (staff training) and 4 (supporting mothers to initiate and maintain breastfeeding) may be particularly important (Fallon et al., 2019).

| Health outcomes
Only two observational studies reported maternal health outcomes.
One study found women delivering in a BFI hospital were less likely to experience mastitis, and the other study showed that women were more likely to remain amenorrhoeic at 6 months when giving birth in a unit with BFI accreditation. Five systematic reviews reported infant 1 year of age. Long-term outcomes up to 6.5 years of age showed increases in neurodevelopment, but no differences in childhood obesity, blood pressure measurements, incidence of allergies or asthma, dental health or child behaviour. The final systematic review (Munn et al., 2016) stated that there was limited evidence from three studies to determine the impact of BFHI on infant health outcomes and that the health outcomes for late preterm infants was not clear.

| Other outcomes
Four reviews reported other outcomes. Maternal knowledge about breastfeeding was higher in women receiving a BFI intervention than a control group and women in BFI intervention groups reported increased breastfeeding support in hospital (Beake et al., 2012). Thematic synthesis of women's experiences across five qualitative studies within one review (Fallon et al., 2019) showed that professional support was highly influential in women's experiences of BFI care; BFI may promote unrealistic expectations and not meet women's individual needs and can have an emotional impact on women especially guilt and feeling pressurised to breastfeed.
Compared to those in control groups, healthcare professionals receiving BFHI training have increased knowledge of the BFI (Beake et al., 2012;Fairbank et al., 2000), were more likely to intend to change their practice (Fairbank et al., 2000) and were more likely to comply with BFHI practices and philosophy within both qualitative and quantitative studies (Munn et al., 2016).

| Subgroup analysis
A subgroup analysis was planned according to income level of the country of the original studies. One review  presented results according to country income level across all included interventions, but not for BFI accreditation specific interventions.
Four reviews with critically low confidence in the findings (Atchan et al., 2013;Chung et al., 2008;Kim et al., 2018 ;Sinha et al., 2017) and one review with high confidence in the findings (Fairbank et al., 2000) either specifically reported the impact of BFI accreditation in low-and middle-income countries (LMIC) or only included BFHI studies from these countries. The reviews, including a total of 14 studies, found BFHI accreditation increased breastfeeding initiation, exclusive or predominant breastfeeding at time points up to 6 months (Atchan et al., 2013;Chung et al., 2008;Fairbank et al., 2000;Kim et al., 2018;Sinha et al., 2017), continued breastfeeding up to 23 months (Sinha et al., 2017) and reduced gastrointestinal infection and atopic dermatitis in infants (Chung et al., 2008). BFHI was viewed to have 'immense potential' to support breastfeeding in LMIC, as could education and counselling interventions (Sinha et al., 2017). However, more research is needed due to limited current evidence (Kim et al., 2018;Sinha et al., 2017), and the currently available data had frequently not been collected for research purposes (Atchan et al., 2013).
Three reviews were either solely focussed on high-income countries or looked at these as a subgroup. Two reviews with critically low confidence in the findings looked at a total of 23 studies from the No reviews reported stage of BFI accreditation, so a subgroup analysis on this aspect could not be performed. a Downgraded (À1) for risk of bias as most included studies for this outcome were medium or high risk of bias and overall confidence in the results of most of the reviews contributing to this outcome was judged to be either low or critically low.

| GRADE summary of findings
b Downgraded (À1) for risk of bias as most included studies for this outcome were of poor methodological design.
c Downgraded (À1) for inconsistency because when considered, heterogeneity was high between studies. d Downgraded (À1) for indirectness because not all included interventions within systematic reviews measured the impact of full BFI accreditation and due to lack of a consistent definition of breastfeeding initiation between included studies. e Downgraded (À1) for imprecision due to lack of clarity over the size of the effect.
f Downgraded (À1) for indirectness because not all included interventions within systematic reviews measured the impact of full BFI accreditation and due to lack of a consistent definition of exclusive breastfeeding between included studies. g Downgraded (À1) for indirectness because not all included interventions within systematic reviews measured the impact of full BFI accreditation and due to lack of a consistent definition of any breastfeeding between included studies.
h Downgraded (À1) for risk of bias as confidence in the findings of the reviews which contributed to this outcome was judged to be crucially low.
i Downgraded (À1) for indirectness as most studies were regarding infants who required neonatal admission or were preterm, few studies looked at health outcomes in all infants.
j Downgraded (À1) for imprecision due to few studies, providing limited evidence regarding the long-term impact of BFI accreditation on infant health.
inconsistency due to heterogeneity being high between studies within the included reviews.

| DISCUSSION
The results of this overview suggest that BFI accreditation may improve breastfeeding initiation and duration of exclusive and any breastfeeding, although no differences were seen within one review investigating longer term breastfeeding between 6 and 23 months.
There is limited current evidence around the impact of BFI on maternal and infant health outcomes. Confidence in this evidence was however judged to be very low for all outcomes.

| Impact of BFI
Breastfeeding intention was not reported within any of the reviews, despite this being a crucial factor in breastfeeding initiation, so the impact of BFI accreditation could not be assessed on this outcome. To achieve BFI accreditation, improved initiation rates have to be evidenced; it is therefore unsurprising that studies showed improved initiation rates, as without this accreditation would not have been achieved (Fallon et al., 2019).
Increasing exclusive breastfeeding is viewed as one of the top interventions to assist in achieving many of the Sustainable Development Goals and to reduce under-5 mortality .
Most reviews found some improvements in exclusivity of breastfeeding; however, the duration of the improvements was much less clear. Similarly, there was lack of clarity over the duration of improvements to the rate of any breastfeeding. Reviews looking at the evidence within high-income countries such as the US and UK questioned the impact of BFI on long-term outcomes (Fallon et al., 2019;Pérez-Escamilla et al., 2016), with community BFI accreditation suggested to lead to changes in the rate of any breastfeeding at 6 to 8 weeks. However, within low-and middle-income settings, improved rates of any breastfeeding with BFI accreditation were seen up to 23 months (Sinha et al., 2017), although the confidence in the findings of this review is critically low.
There was not sufficient evidence reported around long-term maternal health outcomes, and current evidence of the impact of full BFI accreditation on infant health outcomes was based on just one RCT and a few observational studies which showed some impact on reducing gastrointestinal infection and allergic dermatitis, but not other health outcomes. The RCT (Kramer et al., 2001)  impact on long-term health outcomes are essential to allow measurement of any sustained effectiveness of BFI accreditation (Eidelman, 2018).
Studies undertaken in countries with different healthcare systems (e.g., length of hospital stay), economic backgrounds, cultures, rates of breastfeeding initiation or rates of exclusive or any breastfeeding at various time points postpartum may respond differently to BFI accreditation (Fallon et al., , 2019. There is therefore a need to consider the cultural, socioeconomic and practice context of the country where each study occurs in comparison to the country where the impact of BFI is to be assessed.

| Evidence for the individual BFI components
While evaluation of individual BFI steps was not the focus of this overview, it is important to discuss the impact of individual steps, as the causal mechanism for how BFI practices may improve breastfeeding rates have not been fully identified (Munn et al., 2016).
The benefit of some of the individual steps has previously been shown.
Step 4, early skin-to-skin contact, has been shown to increase breastfeeding rates at 1 to 4 months postpartum (Moore et al., 2016).
Community support (part of Step 10) is believed to be essential for long-term breastfeeding duration improvements (Fallon et al., 2019;Pérez-Escamilla et al., 2016) as even in countries with high breastfeeding initiation rates, the drop off of exclusive breastfeeding means countries fall far short of the WHO guidance of 6 months . The impact of Step 3, prenatal breastfeeding education, is unclear with one review stating that prenatal interventions were effective at increasing initiation, exclusivity and duration of breastfeeding although there was lack of clarity over the most effective method of delivery (Wouk et al., 2017) and another review stating there was no conclusive evidence to support antenatal education to improve initiation, exclusivity or duration of breastfeeding (Lumbiganon et al., 2016). Evidence around the impact on

| Cost effectiveness of BFI
The cost of delivering BFI was not an outcome assessed within any of the included systematic reviews. A systematic review that did not meet the criteria for inclusion within this overview (Carroll et al., 2020) however identified three studies that had assessed the cost of implementing BFI at hospital level. All three studies were undertaken in the US. Two studies found that the cost per delivery was increased but that this was not significant, being $35 higher within BFI accredited units than in non-BFI accredited units (DelliFraine et al., 2011) and $19 more in a hospital implementing 6-8 BFI steps than in a hospital implementing 3-5 BFI steps (Allen et al., 2013). However, the first of these studies although matching BFI units to a non-BFI unit within the same city did not match for dep-

| Limitations
The main limitation of this overview is the limited quality of the included reviews and studies. The majority of current evidence is based on before-after studies or cohort studies which did not control for other confounding factors.
It is recognised that not all of the studies incorporated into the included systematic reviews were full BFI accreditation studies with some including studies of individual steps or various combinations of BFI steps. One review also incorporated other 'structured support' interventions alongside BFI interventions.
When judging the quality of included reviews, the authors were only able to assess the information contained within the published articles, alongside some basic searches for review protocols. However, due to word limitations, it is recognised that the published information may not fully reflect all of the processes undertaken within a review.

| Implications
Currently, there is a lack of clear evidence around long-term improved duration of breastfeeding and health benefits of BFI, particularly within high-income countries. Caution is required when determining the potential impact of BFI accreditation in different global contexts using currently available evidence.
Qualitative evidence around BFI has found women to describe that their reality of breastfeeding differed from their expectations and that they can feel pressurised to breastfeed or guilt when unable to succeed in hospitals with BFI accreditation (Fallon et al., 2019).
Woman-centred approaches are increasingly recognised as important, with a need for research to explore the acceptability of the revised 10 steps to parents across a range of international contexts (Aryeetey & Dykes, 2018) and the need to address the growth of cultural acceptance of formula feeding within many countries.

| CONCLUSIONS
This overview suggests that there may be some improvement in initiation and breastfeeding duration from Baby Friendly Initiative accreditation especially in low-income countries, although the duration of any improvements in breastfeeding is uncertain and confidence in these findings was very low due to the poor methodological quality of existing evidence. Evidence around the impact of BFI accreditations on long-term health of mothers and babies is currently minimal. Welldesigned controlled trials are required to better evaluate the shortand long-term impact of BFI accreditation.

ACKNOWLEDGMENTS
Funding for this project was received from Public Health England. The views and opinions expressed within the article are those of the authors and do not necessarily reflect the views Public Health England.

CONFLICTS OF INTEREST
The authors have no competing interests to declare.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analysed in this study.
Did the report of the review contain an explicit statement that the review methods were established prior to the conduct of the review and did the report justify any significant deviations from the protocol? For Partial Yes: The authors state that they had a written protocol or guide that included ALL the following: □ review question(s) □ a search strategy □ inclusion/exclusion criteria □ a risk of bias assessment For Yes: As for partial yes, plus the protocol should be registered and should also have specified: □ a meta-analysis/synthesis plan, if appropriate, □ a plan for investigating causes of heterogeneity □ justification for any deviations from the protocol □ Yes □ Partial Yes □ No 3. Did the review authors explain their selection of the study designs for inclusion in the review? For Yes, the review should provide: □ Explanation for including only RCTs □ OR Explanation for including only NRSI (Non-Randomised Study of Intervention) □ Explanation for including both RCT and NRSI □ Yes □ No 4. Did the review authors use a comprehensive literature search strategy? For Partial Yes (all the following): □ searched at least 2 databases (relevant to research question) □ provided key word and/or search strategy □ justified publication restrictions (e.g. language) For Yes, should also have (all the following): □ searched the reference lists/bibliographies of included studies ( □ searched trial/study registries □ included/consulted content experts in the field □ where relevant, searched for grey literature □ conducted search within 24 months of completion of the review □ Yes □ Partial Yes □ No 5. Did the review authors perform study selection in duplicate? For Yes, either ONE of the following: □ at least two reviewers independently agreed on selection of eligible studies and achieved consensus on which studies to include □ OR two reviewers selected a sample of eligible studies and achieved good agreement (at least 80%), with the remainder selected by one reviewer. □ Yes □ No 6. Did the review authors perform data extraction in duplicate? For Yes, either ONE of the following: □ at least two reviewers achieved consensus on which data to extract from included studies □ OR two reviewers extracted data from a sample of eligible studies and achieved good agreement (at least 80%), with the remainder extracted by one reviewer. □ Yes □ No 7. Did the review authors provide a list of excluded studies and justify the exclusions? For Partial Yes: □ provided a list of all potentially relevant studies that were read in full-text form but excluded from the review For Yes, must also have: □ Justified the exclusion from the review of each potentially relevant study □ Yes □ Partial Yes □ No 8. Did the review authors describe the included studies in adequate detail? For Partial Yes (ALL the following): (Continues) □ described populations □ described interventions □ described comparators □ described outcomes □ described research designs For Yes, should also have ALL the following: □ described population in detail □ described intervention in detail (including doses where relevant) □ described comparator in detail (including doses where relevant) □ described study's setting □ timeframe for follow-up □ yes □ partial yes □ no 9. Did the review authors use a satisfactory technique for assessing the risk of bias (RoB) in individual studies that were included in the review? RCTs For Partial Yes, must have assessed RoB from □ unconcealed allocation □ lack of blinding of patients and assessors when assessing outcomes (unnecessary for objective outcomes such as all cause mortality) For Yes, must also have assessed RoB from: □ allocation sequence that was not truly random, □ selection bias □ Yes □ Partial Yes □ No □ Includes only NRSI NRSI For Partial Yes, must have assessed RoB from □from confounding and □from selection bias For yes, must also have assessed RoB from: □methods used to ascertain exposures and outcomes and □selection of the reported result from among multiple measurements or analyses of the specified outcome □yes □partial yes □no □includes only RCT 10. Did the review authors report on the sources of funding for the studies included in the review? For Yes □ Must have reported on the sources of funding for individual studies included in the review. Note: Reporting that the reviewers looked for this information but it was not reported by study authors also qualifies □ Yes □ No 11. If meta-analysis was performed did the review authors use appropriate methods for statistical combination of results? RCT For Yes: □ The authors justified combining the data in a meta-analysis □ AND they used an appropriate weighted technique to combine study results and adjusted for heterogeneity if present. □ AND investigated the causes of any heterogeneity □ Yes □ No □ No meta-analysis conducted NRSI For Yes: □ The authors justified combining the data in a meta-analysis □ AND they used an appropriate weighted technique to combine study results and adjusted for heterogeneity if present. □ AND they statistically combined effect estimates from NRSI that were adjusted for confounding, rather than combining raw data, or justified raw data when adjusted effect estimates were not available □AND they reported the summary effect estimates for RCTs and NRSI separately when both included in the review □yes □no □no meta-analysis conducted 12. If meta-analysis was performed, did the review authors assess the potential impact of RoB in individual studies on the results of the metaanalysis or other evidence synthesis? For Yes: □ included only low risk of bias RCTs □ OR, if the pooled estimate was based on RCTs and/or NRSI at variable RoB, the authors performed analyses to investigate possible impact of RoB on summary estimates of effect. □ Yes □ No □ No meta-analysis conducted 13. Did the review authors account for RoB in individual studies when interpreting/discussing the results of the review? For Yes: □ included only low risk of bias RCTs □ OR, if RCTs with moderate or high RoB or NRSI were included the review provided a discussion of the likely impact of RoB on the results □ Yes □ No 14. Did the review authors provide a satisfactory explanation for, and discussion of, any heterogeneity observed in the results of the review? For Yes: □ There was no significant heterogeneity in the results □ OR if heterogeneity was present the authors performed an investigation of sources of any heterogeneity in the results and discussed the impact of this on the results of the review □ Yes □ No 15. If they performed quantitative synthesis did the review authors carry out an adequate investigation of publication bias (small study bias) and discuss its likely impact on the results of the review? For Yes: □ performed graphical or statistical tests for publication bias and discussed the likelihood and magnitude of impact of publication bias □ Yes □ No □ No meta-analysis conducted 16. Did the review authors report any potential sources of conflict of interest, including any funding they received for conducting the review? For Yes: □ The authors reported no competing interests OR □ The authors described their funding sources and how they managed potential conflicts of interest □ Yes □ No Note: Shea, B. J., Reeves, B. C., Wells, G., Thuku, M., Hamel, C., Moran, J. … Henry D. A. (2017). AMSTAR 2: A critical appraisal tool for systematic reviews that include randomised or non-randomised studies of healthcare interventions, or both. BMJ, 358, j4008.

Study
Reason Rollins et al., 2016 Discussion around Salera-Vieira & Zembo, 2016 Looks at history of BFI and implementation, not a systematic evaluation of outcomes after BFI accreditation Salt & Romano, 2007 Study is about the implementation of the 'mother-friendly childbirth initiative' for which the BFI ten steps are one element of this initiative. Not a SR on the BFI as an intervention and its impact on BR initiation, duration or exclusivity