key: cord-0967113-hhe2jf3e
authors: Gavine, Anna; Marshall, Joyce; Buchanan, Phyll; Cameron, Joan; Leger, Agnes; Ross, Sam; Murad, Amal; McFadden, Alison
title: Remote provision of breastfeeding support and education: Systematic review and meta‐analysis
date: 2021-12-29
journal: Matern Child Nutr
DOI: 10.1111/mcn.13296
sha: 484ce95e10a41172b9726942acf4a5394d1681d1
doc_id: 967113
cord_uid: hhe2jf3e

The Covid‐19 pandemic has led to a substantial increase in remotely provided maternity care services, including breastfeeding support. It is, therefore, important to understand whether breastfeeding support provided remotely is an effective method of support. To determine if breastfeeding support provided remotely is an effective method of support. A systematic review and meta‐analysis were conducted. Twenty‐nine studies were included in the review and 26 contributed data to the meta‐analysis. Remotely provided breastfeeding support significantly reduced the risk of women stopping exclusive breastfeeding at 3 months by 25% (risk ratio [RR]: 0.75, 95% confidence interval [CI]: 0.63, 0.90). There was no significant difference in the number of women stopping any breastfeeding at 4–8 weeks (RR: 1.10, 95% CI: 0.74, 1.64), 3 months (RR: 0.89, 95% CI: 0.71, 1.11), or 6 months (RR: 0.91, 95% CI: 0.81, 1.03) or the number of women stopping exclusive breastfeeding at 4–8 weeks (RR: 0.86, 95% CI: 0.70, 1.07) or 6 months (RR: 0.93, 95% CI: 0.85, 1.0). There was substantial heterogeneity of interventions in terms of mode of delivery, intensity, and providers. This demonstrates that remote interventions can be effective for improving exclusive breastfeeding at 3 months but the certainty of the evidence is low. Improvements in exclusive breastfeeding at 4–8 weeks and 6 months were only found when studies at high risk of bias were excluded. They are also less likely to be effective for improving any breastfeeding. Remote provision of breastfeeding support and education could be provided when it is not possible to provide face‐to‐face care.

Systematic review evidence shows that support interventions are effective in improving breastfeeding rates (Kim et al., 2018; McFadden et al., 2017 McFadden et al., , 2019 Sinha et al., 2015) . However, support tends to have a greater effect in low and middle countries compared to high income countries (Haroon et al., 2013; . One explanation for this is that within high income countries some form of breastfeeding support is part of routine care and additional support will not confer additional benefits . Moreover, breastfeeding during the pandemic is even more important (Renfrew et al., 2020) . The World Health Organization (WHO) promotes close contact and breastfeeding for mothers and infants affected by COVID-19 as long as mothers practice respiratory hygiene (WHO, 2020) . To date, there is no evidence that COVID-19 is transmitted in breastmilk and the public health gains of breastfeeding outweigh any risks, especially for vulnerable infants (Davanzo et al., 2020; Lackey et al., 2020; WHO, 2020) . Therefore, safe effective support that enables women to initiate and continue breastfeeding and protects healthcare practitioners and lay breastfeeding supporters is imperative.

In the context of COVID-19, remote healthcare provision is essential for service users and staff who are shielding because healthcare settings are perceived as high-risk sources of infection . Remote healthcare provision has the added advantage that healthcare staff who are self-isolating but well, can contribute to service provision without putting themselves or their patients at risk . Variously termed telehealth, e-health and mhealth, remote healthcare provision, including telephone and video consultations, is not new; however, the speed of change during the pandemic is unprecedented (Webster, 2020) . While evidence of clinical effectiveness is equivocal, remote healthcare provision does appear to be acceptable to service users (Ignatowicz et al., 2019; Seuren et al., 2020) . A Cochrane review of health workers experiences of mobile health technologies in primary healthcare reported that staff views were mixed depending on the efficiency of the technologies used and the health workers confidence (Odendaal et al., 2020) . Greenhalgh et al. (2020) suggest that there is considerably more evidence and guidance for telephone consultations.

However, as the aforementioned systematic reviews considered breastfeeding support generally (i.e., peer and professional, face-to-face and remote), there is a need to synthesise evidence via systematic review to understand the effectiveness of interventions provided remotely. In addition to examining how remote support impacts breastfeeding rates it is also important to understand how mothers perceive remote support and if it is associated with changes in perinatal mental health indicators. More specifically, we will include maternal satisfaction with breastfeeding which has been defined as the fulfilment of maternal wishes, expectations or needs in providing breastfeeding for her infant and the pleasure derived from this and may hold more importance for women than duration and exclusivity (Edwards, 2018) .

Maternal self-efficacy is also positively associated with breastfeeding duration and exclusivity (Meedya et al., 2010) , and can be targeted by support interventions. Ensuring these key components are assessed when using remote support strategies serve as a proxy for continued breastfeeding and are important to measure. Perinatal mental health indicators encompass the spectrum of mental health indicators (including and not limited to adjustment disorders and distress, PTSD, mild to moderate depressive illness, severe depressive illness, chronic severe mental illness and postpartum psychosis). Such conditions are associated with adverse breastfeeding outcomes and women could benefit from targeted anticipatory guidance and additional support (Dagla et al., 2021) . In addition as evidence demonstrates the benefits of breastfeeding with reduction in hospitalisations and mortality both in the UK and worldwide, it is important to understand if remote support has any impact on these (Payne & Quigley, 2017; Sankar et al., 2015) .

Finally, we need to understand to what extent remote support can potentially replace face-to-face consultations.

The transition to remote breastfeeding support provision is unlikely to be limited to the context of the COVID-19 pandemic. In the UK, anecdotal evidence before the pandemic, suggested that face-toface breastfeeding support services were being reduced to save money (Better Breastfeeding, 2018) . Support provided remotely may be more cost-effective than face-to-face support although there is currently a lack of evidence to support this within healthcare

• This systematic review investigated the effectiveness of breastfeeding support interventions provided remotely.

• There was significant heterogeneity in how support interventions and standard care were provided.

• There is low-quality evidence that remotely provided breastfeeding support significantly reduced the risk of women stopping exclusive breastfeeding at 3 months but not at 4-8 weeks or 6 months. There was no significant difference in the number of women stopping any breastfeeding.

• Interventions tended to demonstrate more positive effects when standard care was limited suggesting remote support is preferable to no support.

• There was a lack of evidence exploring women's satisfaction and the impact on maternal mental health.

More research is needed to explore this.

generally (Gonçalves-Bradley et al., 2020) . However, there may be some advantages to remote breastfeeding support such as improved accessibility for those living in remote and rural areas , and improved continuity of care (Friesen et al., 2015) . It is therefore important to understand whether breastfeeding support provided remotely is effective.

To inform care during and beyond the COVID-19 pandemic, we conducted a systematic review and meta-analysis to determine the most effective methods of providing breastfeeding support and education remotely to pregnant and breastfeeding women. The objectives of the review were to determine the effectiveness of different modes of remote provision of breastfeeding support and education; different providers of remote breastfeeding support and education; and different timings and frequencies of remote breastfeeding support and education on breastfeeding rates and also on the following secondary outcomes: maternal satisfaction; perinatal mental health indicators; maternal self-efficacy; infant and child morbidity and mortality and number of women requiring additional face-to-face support.

The protocol for this systematic review is registered on PROSPERO (CRD42020176130).

We searched five electronic databases: MEDLINE, CINAHL, MIDIRS, CENTRAL and EMBASE. Searches were conducted in March 2020 using a combination of index and free-text terms related to 'breastfeeding' AND 'remote care' AND trials (for full search strategy see Supporting Information File S1). No language limit was placed on the search, however, only studies reported in English were included. Studies published before 2010 were excluded to ensure the included interventions reflected the most up-to-date technologies. We also scanned the reference lists of all included studies and relevant systematic reviews.

Studies were included if they were individual or cluster-randomised trials. Studies were eligible for inclusion if they contained the following groups of women: pregnant women intending to breastfeed; mothers who may initiate breastfeeding; mothers who are breastfeeding; mothers who ceased breastfeeding who wished to re-start.

This included women who have specific health problems and mothers of preterm infants or infants requiring additional medical care. Women who gave birth via Caesarean Section or vaginal birth were included. Studies were included if the intervention was also aimed at fathers and/or other caregivers as well as mothers.

Interventions were eligible for inclusion if they involved the provision of breastfeeding support or education provided remotely, defined as where service users and service-providers are separated by distance, for example, telephone, text messaging, social media, video call and e-mail (WHO, 2016) . Studies were included if the intervention occurred in the antenatal or postnatal periods alone or in both the antenatal and postnatal periods. Interventions that also contained a face-to-face element were only included if this was delivered post-partum and before discharge home and was part of usual care and thus also received by the control group.

Interventions could be offered by health professionals or lay people or both, trained or untrained. It could be offered to groups of women or one-to-one and it could be offered proactively by contacting women directly, or reactively, by waiting for women to get in touch.

'Support' interventions eligible for this review could include elements such as reassurance, praise, information, and the opportunity to discuss and to respond to the mother's questions. Interventions had to be two-way interactions between the supporter and participant. It could include discussing the practical management of breastfeeding (e.g., attachment of the infant, identifying an infant's feeding cues). Education interventions eligible for this review had to be provided to the mother and may or may not also have included the father or other caregivers.

The comparator was either standard care, no breastfeeding support, or education or breastfeeding support provided face-to-face.

To be included studies had to report one of the following primary outcomes:

1. Number of women who stop exclusive breastfeeding at 4-8 weeks.

2. Number of women who stop any breastfeeding at 4-8 weeks.

3. Number of women who stop exclusive breastfeeding at 3 months. 4. Number of women who stop any breastfeeding at 3 months. 5. Number of women who stop exclusive breastfeeding at 6 months. 6. Number of women who stop any breastfeeding at 6 months.

The outcomes focused on the number of women who stopped breastfeeding to allow for easier comparison with the Cochrane review on Breastfeeding Support (McFadden et al., 2017) .

The following secondary outcomes were also included:

1. Maternal satisfaction.

3. Maternal self-efficacy. 4. Infant and child morbidity and mortality. 5. Number of women requiring additional face-to-face support.

Studies were excluded if the intervention was targeted at individuals or organisations providing breastfeeding support or education and did not collect data on maternal or infant outcomes. Studies that have a face-to-face element beyond providing support or education before discharge after birth were excluded.

Interventions that provided only one-way information from a care provider were excluded (e.g., text message reminders).

Titles, abstracts and potentially relevant full texts were screened independently by two authors against the eligibility criteria. Any disagreement was resolved through discussion and consultation with a third author.

Two authors independently extracted information on participants, intervention, context and outcomes using a specifically designed data extraction form. Any discrepancies were resolved through discussion.

When information regarding study methods and results was unclear, we attempted to contact authors to provide further details. Two authors independently assessed risk of bias for each study using the criteria outlined in the Cochrane Handbook for systematic reviews of Interventions . We assessed the quality of the evidence using the GRADE approach (Schünemann et al., 2013) , with the GRADEpro online programme (https://gradepro.org/).

Given the pressing need to complete this study in a rapid timeframe to inform how breastfeeding support can be provided in the Covid-19 situation, primary outcomes were prioritised. Meta-analysis of the primary outcomes was performed using Review Manager 5 software.

The primary outcomes are presented as summary risk ratios with 95% confidence intervals. We anticipated heterogeneity between studies in terms of the interventions and populations, so a random-effects meta-analysis was used .

Where possible we aimed to include all available data for all randomised patients, in the group to which they were randomised.

We contacted authors regarding any missing data. Where data were missing, we assumed that the woman had ceased breastfeeding. Thus the denominator for each outcome in each trial was the number randomised minus any participants whose outcomes were known to be missing.

The sample sizes of cluster randomised trials were adjusted using the intracluster correlation coefficient when possible. If these were not presented, we contacted authors for this information. Heterogeneity was quantified using the I 2 statistic.

Secondary outcomes are described narratively and used to aid understanding of the impact of the intervention on primary outcomes.

Cluster randomized trials were adjusted for clustering using the formula detailed in the Cochrane Handbook (Higgins et al., 2019) . First the design effect was calculated using the following formula: 1 + (average cluster size -1) × intracluster correlation coefficient (ICC). The effective sample size for the intervention and control group was then calculated by dividing the number of participants in each group by the design effect. All studies reported the ICC.

We carried out sensitivity analyses on missing data and trial quality.

For missing data, we excluded studies with high (>20% attrition) or unclear risk of attrition bias. For quality, we excluded all studies with high or unclear risk of bias in the allocation concealment domain of the Cochrane Risk of Bias tool.

We assessed the risk of publication bias through a visual inspection of Funnel Plots for each analysis. To assess whether any asymmetry was a result of publication bias, the effect estimates from the random-effects model was compared with a fixed-effects model. If the random effects model was found to increase the effect estimate, one explanation could be the presence of publication bias Sterne et al., 2011) .

The search identified 3188 original records. One further paper was identified by searching reference lists of included papers and one further paper was published after the search and identified by searching for papers related to relevant protocols. On titles and abstracts screening, 3080 papers were excluded. Following review of 110 full texts articles, 61 were excluded. This left 49 papers reporting 29 studies included in this review (see Figure 1 ). To facilitate ease of reading we have included only the primary reference from each study within the manuscript text. The additional references are detailed inTable S1. Three studies did not report relevant outcomes or did not present data usable in the meta-analysis Niela-Vilen et al., 2016; Reeder et al., 2014) and

were therefore not included in the analysis; 26 studies contributed data to the meta-analysis.

The 29 studies included one cluster randomised trial (Fu et al., 2014) , and 28 individually randomised trials. Twenty-five were 2 arm trials with 4 having 3 arms (Fu et al., 2014; Kamau-Mbuthia et al., 2013; Reeder et al., 2014; Unger et al., 2018) . In 3 of these studies only one intervention arm was relevant and included in this review (Fu et al., 2014; Kamau-Mbuthia et al., 2013; Unger et al., 2018) . In Reeder et al. both intervention arms were relevant, but data were not presented in a form that could be included in a meta-analysis. Specifically, Fu et al. had one intervention arm with an educational intervention delivered solely in hospital before discharge (excluded) and another with support and education provided by telephone (included); Kamau-Mbuthia et al. had one arm delivered face-to-face peer groups (excluded) and one telephone support (included); and Unger et al. had one intervention weekly one way texts and another weekly two way text messaging (two way only included).

Full detail of included studies and description of interventions and comparisons is provided in Table 1 .

A total of 11,470 women from 29 trials (sample sizes ranged from 65 to 1948) and their infants from 17 countries were randomised.

Twenty-three studies were conducted in 12 high-income countries accounting for 79% of participants (9117 participants). Three studies were conducted in upper-and 3 in lower-middle income countries accounting for 6.8% (786) and 13.6% (1567) participants, respectively. No studies were identified that had been conducted in lowincome countries.

Characteristics of participants varied across the studies both for mothers and infants. Some studies recruited from specific populations groups, such as adolescent mothers (Di Meglio et al., 2010) , obese mothers (Carlsen et al., 2013) , or low income mothers (Efrat et al., 2015; Hoddinott et al., 2012; Kamau-Mbuthia et al., Palacios et al., 2018; Patel et al., 2018; Unger et al., 2018) . Interactive text messaging alone was used in four studies Efrat et al., 2015; Lucas et al., 2019; Martinez-Brockman et al., 2018; Palacios et al., 2018; Unger et al., 2018) and with hyperlinks to an educational module in one study (Lucas et al., 2019) . A phone App was used in four studies (Ahmed et al., 2016; Hagi-Pedersen et al., 2017; Lewkowitz et al., 2020; Uscher-Pines et al., 2019) . Other interventions included: peer support via social media (e.g., Facebook) either in response to mothers concerns (Niela-Vilen et al., 2016) , or structured by use of an educational booklet (Cavalcanti et al., 2019) ; video calls or Skype (Seguranyes et al., 2014) ; support via a website alone (Gonzalez-Darias et al., 2020) or via a website combined with other interventions such as a workbook, and a DVD with telephone calls (Abbass-Dick et al., 2015) .

All interventions were provided after birth but in seven studies there was also an antenatal component Efrat et al., 2015; Kamau-Mbuthia et al., 2013; Martinez-Brockman et al., 2018; Patel et al., 2018; Reeder et al., 2014; Unger et al., 2018) . The frequency of interventions was variable with nine being open access and/or whenever needed by the mother (Ahmed et al., 2016; Gonzalez-Darias et al., 2020; Kamau-Mbuthia et al., 2013; Lewkowitz et al., 2020; Maslowsky et al., 2016; Niela-Vilen et al., 2016; Seguranyes et al., 2014; Simonetti et al., 2012; Uscher-Pines et al., 2019) . In seven studies interventions were provided weekly (Abbass-Dick et al., 2015; Cavalcanti et al., 2019; Forster et al., 2019; Fu et al., 2014; Palacios et al., 2018; Reeder et al., 2014; Unger et al., 2018) and six were every 1-2 days Ericson et al., 2018; Hagi-Pedersen et al., 2017; Hoddinott et al., 2012; Lucas et al., 2019; Patel et al., 2018) . Six varied in frequency (Carlsen et al., 2013; Di Meglio et al., 2010; Efrat et al., 2015; Martinez-Brockman et al., 2018; Reeder et al., 2014; Simonetti et al., 2012) , one was twice monthly (Tahir & Al-Sadat, 2013) , one was only provided once (Howell et al., 2014) , and in one study frequency was not specified (Hongo et al., 2020) . A range of personnel provided the interventions. In 11 studies interventions were provided by health care professionals, either nurses or midwives with or without specialist training (Cavalcanti et al., 2019; Ericson et al., 2018; Hagi-Pedersen et al., 2017; Hoddinott et al., 2012; Lucas et al., 2019; Maslowsky et al., 2016; Palacios et al., 2018; Patel et al., 2018; Seguranyes et al., 2014; Simonetti et al., 2012; Unger et al., 2018) and in one study by social workers (Howell et al., 2014) . In eight studies interventions were provided by lactation consultants, although it was not always clear what specialist training had been completed (Abbass-Dick et al., 2015; Ahmed et al., 2016; Carlsen et al., 2013; Demirci et al., 2019; Fu et al., 2014; Lewkowitz et al., 2020; Tahir & Al-Sadat, 2013; Uscher-Pines et al., 2019) .

In a further nine studies interventions were provided by lay or peer supporters (Di Meglio et al., 2010; Efrat et al., 2015; Forster et al., 2019; Gonzalez-Darias et al., 2020; Hongo et al., 2020; Kamau-Mbuthia et al., 2013; Martinez-Brockman et al., 2018; Niela-Vilen et al., 2016; Reeder et al., 2014) and a midwife was also available in one study (Niela-Vilen et al., 2016) .

Most Martinez -Brockman et al., 2018; Reeder et al., 2014; Tahir & Al-Sadat, 2013; Uscher-Pines et al., 2019) , two included the option to call the healthcare provider for advice (Ericson et al., 2018; Hoddinott et al., 2012) and three included both (Ahmed et al., 2016; Forster et al., 2019; Simonetti et al., 2012) . In six studies the comparison was described as routine postnatal care with no information about breastfeeding support (Gonzalez-Darias et al., 2020; Maslowsky et al., 2016; Niela-Vilen et al., 2016; Patel et al., 2018; Seguranyes et al., 2014; Unger et al., 2018) . Three studies described combined support and breastfeeding education (Ahmed et al., 2016; Cavalcanti et al., 2019; Efrat et al., 2015) . The length of time standard care was provided was reported in 12 studies and varied. It could be as needed (Abbass-Dick et al., 2015; Cavalcanti et al., 2019) , last several weeks (Ahmed et al., 2016; Carlsen et al., 2013; Ericson et al., 2018; Forster et al., 2019; Hoddinott et al., 2012; Howell et al., 2014; Lucas et al., 2019; Maslowsky et al., 2016; Seguranyes et al., 2014) or months after discharge (Simonetti et al., 2012) . perinatal mental health and one study measured maternal selfefficacy. Five studies only measured breastfeeding rates at 3 months and a post-hoc decision was made to include this as an outcome to allow these studies to contribute data to the meta-analyses.

The risk of bias varied across the domains. More specifically, 75% of studies were judged as low risk of bias for random sequence generation; 48% were low risk of bias for allocation concealment; 59%

were low risk of bias for incomplete outcome data; 55% were low risk of bias for selective outcome reporting and 83% were low risk for any other forms of bias. Reasons for high risk of bias in the 'other' domain was generally due to significant baseline imbalances. All studies were judged to be at high or unclear risk of performance bias due to the difficulties in blinding participants and personnel to the interventions.

Except for Patel et al. (2018) , all studies were judged to be either high or unclear risk of bias for blinding of outcome assessment. This was a consequence of most studies collecting self-report data from the women who were not blinded. See Figures S1,2 for further details.

Primary outcomes Sensitivity analyses which removed the 13 studies at high or unclear risk of bias for allocation concealment found similar effects (see | 15 of 23 sensitivity analysis whereby the 12 studies at high or unclear risk of bias for incomplete outcome data (see Table S2 ), found that exclusive breastfeeding at 4-8 weeks became significant with a 24% reduction in risk of stopping breastfeeding (RR: 0.76, 95% CI: 0.61, 0.96). Similarly, any breastfeeding at 3 and 6 months just reached statistical significance with a reduced risk in stopping breastfeeding of 26% (RR: 0.74, 95% CI: 0.55, 1.00) and 18% (RR: 0.82, 95% CI: 0.67, 1.00), respectively.

Inspection of funnel plots did identify some asymmetry with smaller studies tending to show more positive effects (see Figures S1-S6). However, there is also evidence of substantial heterogeneity which may contribute to this. To further explore the risk of publication bias, the fixed and random effect sizes were compared.

The random effects model did not shift the effect estimates towards the results of the smaller studies which suggests an absence of any small study effects Sterne et al., 2011) .

Three of the four studies measuring maternal satisfaction reported no differences between the study groups. Ericson et al. (2018) However Patel et al. (2018) , reported that 92.3% of the intervention group were satisfied versus 36% of the control group.

Only one study reported measures relating to perinatal mental health (Ericson et al., 2018) . There was a small but significant im- 

The six primary outcomes were assessed with the GRADE criteria (see Table 2 ). All outcomes were graded low or very low. Outcomes were not downgraded for lack of blinding. Although Funnel Plot asymmetry was identified we did not downgrade for this as a F I G U R E 6 Forest plot of comparison: Remote support versus standard care/control, outcome: Stopping exclusive breastfeeding at 6 months F I G U R E 7 Forest plot of comparison: Remote support versus standard care/control, outcome: Stopping exclusive breastfeeding at 6 months comparison with a fixed effects model did not reveal that the random effects estimate was more beneficial.

This systematic review found that remotely provided breastfeeding support and education combined with support in hospital is an effective intervention to increase the rates of exclusive breastfeeding at 3 months. However, for other outcomes the results are less clear.

Exclusive breastfeeding at 6 months was only significant when studies at high risk of allocation concealment were excluded. Exclusive breastfeeding at 4-8 weeks was only significant once studies with loss to follow-up greater than 20% were excluded. There was little evidence of any impact on any breastfeeding. However, when studies with loss to follow-up greater than 20% were excluded, any breastfeeding at 3 and 6 months did become significant. The certainty of the evidence was judged to be low or very low for all outcomes.

While this may suggest that remote breastfeeding support and education had little effect on duration of any breastfeeding, it is important to consider what the interventions were being compared to, with the majority of studies providing some form of breastfeeding support as part of their routine care. There was considerable heterogeneity in how routine care was offered (e.g., lactation consultants, peer support, in-hospital, outpatient clinics, reactive or proactive), which makes comparisons difficult. Moreover, in the studies reporting statistically significant benefits, the control group appeared to have more limited support, e.g. offered reactively or only limited details provided (Abbass-Dick et al., 2015; Cavalcanti et al., 2019; Gonzalez-Darias et al., 2020; Hoddinott et al., 2012; Kamau-Mbuthia et al., 2013; Simonetti et al., 2012; Tahir & Al-Sadat, 2013) . It therefore could be argued that if remote provision is the only type of provision available (as would be the case in the Covid-19 pandemic), then this may have been more effective than the comparator. Alternatively, if remote provision in addition to existing services including face-to-face support is synergistic then inability to provide these may reduce the effectiveness of the remote provision.

Given the low number of studies and high heterogeneity, it is difficult to extrapolate from our review, why remote support has a greater effect for exclusive breastfeeding compared to any breastfeeding and why this occurs only at certain timepoints. One reason could be that mothers who are highly motivated to breastfeed exclusively benefited most from support to achieve their goals. This explanation is supported by a realist review of breastfeeding peer support (Trickey et al., 2018) Other systematic reviews have also found that breastfeeding support has a greater effect on exclusive compared to any breastfeeding outcomes (McFadden et al., 2017; McFadden et al., 2019) . In addition, McFadden et al. (2017) found that support was more effective in settings where background rates of breastfeeding initiation were high. This suggests that continued efforts are needed to promote breastfeeding to increase motivation.

Explanations for why support interventions were effective for exclusive breastfeeding at 3 but not 6 months are probably multi-T A B L E 2 Summary of findings faceted. Overall, the breastfeeding interventions tended to have higher intensity in the first 6 weeks following birth, with frequency of contact reducing or ceasing beyond this timepoint. Reasons why mothers cease breastfeeding in the first 6 weeks such as fatigue,

inconvenience and concerns about milk supply (C. R. L. Brown et al., 2014; McAndrew et al., 2012) , may be more amenable to the support interventions There are additional contextual challenges to exclusive breastfeeding beyond 3 months, for example mothers returning to paid employment, cultural practices of introducing solids before 6 months and societal attitudes towards breastfeeding public in some settings (Alianmoghaddam et al., 2018; Burns & Triandafilidis, 2019) . These challenges may impact on the effectiveness of support interventions.

In addition, we cannot extrapolate whether any specific participant characteristics such as age, level of education or parity impacted on the effectiveness of interventions. These characteristics are not consistently reported in trials and their association with breastfeeding varies in different settings.

As expected, statistical heterogeneity was high for all primary outcomes. Due to the relatively small number of studies and unequal distribution of the co-variates, formal sub-group analysis was not undertaken. However, a qualitative examination of the characteristics of the interventions indicated that the studies which reported positive intervention effects did not differ greatly from those that reported no significant differences or negative effects in terms of who provided the intervention, whether it was provided one-to-one or to groups or whether it was provided postnatally only or antenatally and postnatally. However, it is worth noting that only nine studies reported a positive effect on at least one breastfeeding outcome and of these seven used the telephone as a mode of support (Cavalcanti et al., 2019; Di Meglio et al., 2010; Hoddinott et al., 2012; Kamau-Mbuthia et al., 2013; Patel et al., 2018; Simonetti et al., 2012; Tahir & Al-Sadat, 2013) . This may suggest that telephone is a more effective way of providing support, however, results should be interpreted with caution due to the risk of bias in these studies and the lack of formal sub-group analysis.

The studies that did not use telephone calls used other delivery methods including video conferencing, SMS and/or platforms to facilitate online text-based interventions. Current evidence for health interventions delivered using these platforms have only shown effectiveness in specific groups of participants, for example, those with stable chronic illnesses (Free et al., 2013; Greenhalgh et al., 2020) , and therefore may not be applicable to breastfeeding women.

Nevertheless in Seguranyes et al. (2014) , 90% of women chose videoconferencing over telephone calls, although only 40% of the intervention group made use of any form of teleconsultation. Over 80% of women who did not use services said they had no reason to do so, which suggests it is not the mode of the delivery that impacted on uptake of the intervention and for those seeking help, videoconferencing was the preferred option.

Four studies measured the secondary outcome of maternal satisfaction, with only one of these reporting significantly higher scores in women receiving remote support (Patel et al., 2018) . This could be explained by the fact that, in the three studies where no significant difference was identified, women in the control group were provided either with reactive breastfeeding support (Ericson et al., 2018; Hoddinott et al., 2012) or face-to-face support (Seguranyes et al., 2014) . Whereas women in the control group in the study by Patel et al. (2018) , were not provided breastfeeding support. This may therefore suggest that any form of breastfeeding support is perceived positively by women. This is consistent with findings from an integrative review on telehealth and breastfeeding (Ferraz dos Santos et al., 2020) . There was insufficient information to draw conclusions about the other secondary outcomes of perinatal mental health and maternal self-efficacy as few studies measured these outcomes.

Overall, this review found less evidence of positive intervention effects than existing reviews which considered on support more generally (Haroon et al., 2013; McFadden et al., 2017; McFadden et al., 2019; Olufunlayo et al., 2018; Shakya et al., 2017) . Furthermore, reviews that have specifically assessed the effect of telephone in relation to face-to-face support via sub-group analysis have identified that face-to-face support appears to be more effective than telephone support alone, although caution of interpretation is were generally of a lower intensity in terms of numbers of contacts and the web-based ones required the mother to reactively access it.

Previous evidence suggests that generally higher intensity delivered proactively may be more successful (McFadden et al., 2017; Trickey et al., 2018) . Moreover, a more nuanced approach whereby a schedule of contacts is agreed between the mother and supporter may better meet the needs of women (Trickey et al., 2018) .

Breastfeeding support is complex and there may be important elements that are not easily addressed remotely. In addition to practical and informational elements, support incorporates emotional support and esteem-building as well as social support, such as signposting women to support groups and helping to build their social networks. Although it is possible to offer emotional support via telephone or using other remote technologies this may be more challenging and may require specific training for supporters (Penny et al., 2018 (Penny et al., 2018) , more recent research has demonstrated an improvement in the capacity of staff to provide digital support (Fortuna et al., 2020) . It is therefore plausible, that remotely provided interventions provided in a post-Covid context, may be more effectively implemented due to higher skill levels of providers. Such interventions offer greater flexibility for the user, allowing for choice, and individual preference (e.g., a text-based intervention may be more useful if a mother is caring for a crying infant). Given the increasing range of digital platforms, there is a need for qualitative work to explore women's experiences with them and that could provide more insights into why such interventions succeed or fail.

There is a lack of evidence relating to low-and middle-income countries. No studies from low-income countries were included in this review and only 6 from middle-income countries (20% of participants).

Most interventions in this review involved the use of a telephone (e.g., calls, texts) or smartphone (e.g., websites, Facebook, apps) and while ownership may be affected by poverty, surveys have shown that phone ownership is relatively high in some low-income countries, although this varies greatly (Taylor & Silver, 2019) . Additionally, there is also inequity in access within high-income countries, with families on a lower income and those with lower educational attainment less likely to have access to a smartphone and home broadband (Katzow et al., 2020) . It may be that the provision of remote support increases accessibility for some mothers (e.g., remote and rural) and can be offered out of regular office hours covering evening, weekends and holidays but some mothers may be excluded by lack of access to technologies or by costs of phone and/or internet access.

Finally, there was a lack of evidence on support for women with preterm infants. As additional considerations are needed for preterm infants, for example kangaroo care (Charpak et al., 2021) responsive versus scheduled feeding (Watson & McGuire, 2016) , the use of milk banks (Quigley et al., 2019) , use of bottles or cups (Collins et al., 2016) , and therefore the evidence from studies with term babies may not be applicable.

A key strength of this systematic review was that it followed the methodology outlined in the Cochrane Handbook (Higgins et al., 2019) and the GRADE approach was used to help judge the certainty of the evidence (Schünemann et al., 2013) . This review had some important limitations. Standard care for those in the control groups was not well-described or when described it was variable. In many studies face-to-face provision was part of standard care and this may not have been possible during lockdowns.

A second limitation of this review is the high drop-out rate, with 11 studies being rated as high risk of bias and 2 as unclear risk of bias for this domain. In our intention-to-treat analysis we assumed all women did not complete follow-up had stopped breastfeeding. However, this potentially leads to an underestimation of effect size as a sensitivity analysis identified that differences in exclusive breastfeeding at 4-8 weeks and any at 6 months became significant when studies at high or unclear risk of bias in this domain were excluded. A qualitative examination of the studies did not identify any clear differences in attrition rates based on mode of delivery.

Third, a post-hoc decision was made to include any or exclusive breastfeeding at 3 months as an outcome. This decision was made because several studies otherwise meeting the eligibility criteria only measured breastfeeding rates at this time point. Fourth, statistical heterogeneity was high for all outcomes and is likely a manifestation of heterogeneity in participants and interventions. Fifth, there was some evidence of funnel plot asymmetry, however, due to high levels of heterogeneity we cannot be certain that this was due to publication bias.

Finally, only studies published in English were eligible for inclusion. One full-text paper which may have potentially met the inclusion criteria was excluded for this reason (Araque García et al., 2018) , and others may have been excluded at the title and abstract screening stage.

The findings of this systematic review and meta-analysis demonstrate that remote interventions can be effective for improving exclusive breastfeeding but only at specific time frames. Remote interventions were effective at increasing exclusive breastfeeding at 3 months but at 4-8 weeks and 6 months they were effective only when studies at high risk of bias were removed in a sensitivity analysis. However, remotely provided breastfeeding support was less likely to be effective for improving any breastfeeding and only reached significance at 3 and 6 months when studies with rates of attrition were excluded. Moreover, the certainty of the evidence was judged to be low or very low due to risk GAVINE ET AL.

| 19 of 23 of bias, substantial heterogeneity and imprecision in some outcomes.

High levels of attrition were an issue in 11 of 26 studies and when these studies were removed differences in any breastfeeding at 4-8 weeks and 6 months became significant. There was significant heterogeneity in terms of interventions and routine care making comparisons difficult, however, interventions tended to show more positive effects when only limited routine care was provided. Given that the need for breastfeeding support is highlighted in the wider literature, this review suggests that remote provision of breastfeeding support and education should be provided when it is unsafe or not possible to provide face-to-face care.

Beyond the Covid-19 pandemic, remote support could be offered as part of a support package that could supplement face-to-face provision. This would offer women more flexibility to access help when needed to meet individual needs and changing circumstances. To inform policy and longer-term provision globally there is a need to further understand the use of contemporary technologies and women's and care providers' experiences of using these.

Maternal and paternal experiences and satisfaction with a co-parenting breastfeeding support intervention in Canada

Coparenting breastfeeding support and exclusive breastfeeding: A randomized controlled trial

The effect of interactive web-based monitoring on breastfeeding exclusivity, intensity, and duration in healthy, term infants after hospital discharge

Reasons for stopping exclusive breastfeeding between three and six months: A qualitative study

Efectividad de una consulta online de Enfermería en el seguimiento e instauración de la lactancia materna

Cuts to breastfeeding support in England

Breastfeeding and childhood acute otitis media: A systematic review and meta-analysis

Experiences of breastfeeding during COVID-19: Lessons for future practical and emotional support

Factors influencing the reasons why mothers stop breastfeeding

Taking the path of least resistance: A qualitative analysis of return to work or study while breastfeeding

Telephone-based support prolongs breastfeeding duration in obese women: A randomized trial

Online participatory intervention to promote and support exclusive breastfeeding: Randomized clinical trial

Systematic review and meta-analysis suggest that the duration of Kangaroo mother care has a direct impact on neonatal growth

Breastfeeding and maternal health outcomes: A systematic review and meta-analysis

Avoidance of bottles during the establishment of breast feeds in preterm infants

Women's mental health as a factor associated with exclusive breastfeeding and breastfeeding duration: Data from a longitudinal study in Greece

Breastfeeding and coronavirus disease-2019: Ad interim indications of the Italian Society of Neonatology endorsed by the Union of

Position statement on breastfeeding from the Italian pediatric societies

Cochrane Handbook for Systematic Reviews of Interventions Version

Design, implementation, and feasibility-acceptability of a text message-based intervention to prevent perceived insufficient milk supply

A randomized controlled trial of telephone peer support's influence on breastfeeding duration in adolescent mothers

An exploration of maternal satisfaction with breastfeeding as a clinically relevant measure of breastfeeding success

The effect of lactation educators implementing a telephone-based intervention among low-income Hispanics: A randomised trial

Breastfeeding and risk for ceasing in mothers of preterm infants: Long-term follow-up

Proactive peer (mother-to-mother) breastfeeding support by telephone (ringing up about breastfeeding early

Strategies to increase peer support specialists' capacity to use digital technology in the era of COVID-19: Pre-post study

The effectiveness of mobile-health technologybased health behaviour change or disease management interventions for health care consumers: A systematic review

Using videoconferencing technology to provide breastfeeding support to low-income women: Connecting hospital-based lactation consultants with clients receiving care at a community health center

Professional breastfeeding support for first-time mothers: A multicentre cluster randomised controlled trial

Mobile technologies to support healthcare provider to healthcare provider communication and management of care

Supporting a first-time mother': Assessment of success of a breastfeeding promotion programme

Video consultations for covid-19: An opportunity in a crisis

Multicentre randomised study of the effect and experience of an early inhome programme (PreHomeCare) for preterm infants using video consultation and smartphone applications compared with inhospital consultations: Protocol of the PreHomeCare study

Breastfeeding promotion interventions and breastfeeding practices: A systematic review

Cochrane handbook for systematic reviews of interventions version 5

The Cochrane Collaboration

The FEeding Support Team (FEST) randomised, controlled feasibility trial of proactive and reactive telephone support for breastfeeding women living in disadvantaged areas

The influence of breastfeeding peer support on breastfeeding satisfaction among Japanese mothers: A randomized controlled trial

Breastfeeding and intelligence: A systematic review and meta-analysis

Long-term consequences of breastfeeding on cholesterol, obesity, systolic blood pressure and type 2 diabetes: A systematic review and metaanalysis

Short-term effects of breastfeeding: A systematic review on the benefits of breastfeeding on diarrhoea and pneumonia mortality (9241506121)

An intervention to extend breastfeeding among black and Latina mothers after delivery

Internet videoconferencing for patient-clinician consultations in long-term conditions: A review of reviews and applications in line with guidelines and recommendations

Effect of a peer support service on breast-feeding continuation in the UK: A randomised controlled trial

Systematic review of peer support for breastfeeding continuation: Metaregression analysis of the effect of setting, intensity, and timing

The Australian Breastfeeding Association's national breastfeeding helpline: A report of the experiences of callers and volunteers in 2019

Cell phone based peer counseling to support exclusive breastfeeding is associated with more frequent help and decreased breastfeeding problems: O079

The use of and experiences with telelactation among rural breastfeeding mothers: Secondary analysis of a randomized controlled trial

Telemedicine and health disparities during COVID-19

Interventions promoting exclusive breastfeeding up to six months after birth: A systematic review and meta-analysis of randomized controlled trials

SARS-CoV-2 and human milk: What is the evidence? Maternal & Child Nutrition

Impact of a novel smartphone application on low-income women's breastfeeding rates: A randomized controlled trial

Parents' experiences of using videoconferencing as a support in early discharge after childbirth

Efficacy of a breastfeeding pain self-management intervention: A pilot randomized controlled trial

Impact of the lactation advice through texting can help (LATCH) trial on time to first contact and exclusive breastfeeding among WIC participants

Effects of postpartum mobile phone-based education on maternal and infant health in Ecuador

Infant feeding survey

Support for healthy breastfeeding mothers with healthy term babies

Counselling interventions to enable women to initiate and continue breastfeeding: A systematic review and meta-analysis

Factors that positively influence breastfeeding duration to 6 months: A literature review

Breastfeeding preterm infants-A randomized controlled trial of the effectiveness of an Internet-based peer-support group

Health workers' perceptions and experiences of using mHealth technologies to deliver primary healthcare services: A qualitative evidence synthesis

Interventions to improve breastfeeding exclusivity in low-and middle-income countries: A systematic review and meta-analysis

Chapter 13: Assessing risk of bias due to missing results in a synthesis

Effect of a multi-site trial using short message service (SMS) on infant feeding practices and weight gain in low-income minorities

Effectiveness of weekly cell phone counselling calls and daily text messages to improve breastfeeding indicators

Breastfeeding and infant hospitalisation: Analysis of the UK

Registered nurse and midwife experiences of using videoconferencing in practice: A systematic review of qualitative studies

Effect of breastfeeding on malocclusions: A systematic review and meta-analysis

Formula versus donor breast milk for feeding preterm or low birth weight infants

Guidance for antenatal and postnatal services in the evolving coronavirus (COVID-19) pandemic

Telephone peer counseling of breastfeeding among WIC participants: A randomized controlled trial

Optimising mother-baby contact and infant feeding in a pandemic: Rapid analytic review

Why invest, and what it will take to improve breastfeeding practices? The Lancet

Optimal breastfeeding practices and infant and child mortality: A systematic review and meta-analysis

GRADE handbook for grading quality of evidence and strength of recommendations

Efficacy of a videoconferencing intervention compared with standard postnatal care at primary care health centres in Catalonia. Midwifery

Physical examinations via video? Qualitative study of video examinations in heart failure, using conversation analysis

Effectiveness of community-based peer support for mothers to improve their breastfeeding practices: A systematic review and meta-analysis

A structured telephonic counselling to promote the exclusive breastfeeding of healthy babies aged zero to six months: A pilot study

Interventions to improve breastfeeding outcomes: A systematic review and meta-analysis

Recommendations for examining and interpreting funnel plot asymmetry in metaanalyses of randomised controlled trials

Does telephone lactation counselling improve breastfeeding practices? A randomised controlled trial

Smartphone ownership is growing rapidly around the World, but not always equally

Breastfeeding and the risk of dental caries: A systematic review and meta-analysis

UK Breastfeeding Helpline support: An investigation of influences upon satisfaction

A realist review of one-to-one breastfeeding peer support experiments conducted in developed country settings

Short message service communication improves exclusive breastfeeding and early postpartum contraception in a low-to middle-income country setting: A randomised trial

Feasibility and effectiveness of telelactation among rural breastfeeding women

Breastfeeding in the 21st century: Epidemiology, mechanisms, and lifelong effect

Responsive versus scheduled feeding for preterm infants

Virtual health care in the era of COVID-19

Global diffusion of eHealth: Making universal health coverage achievable

Remote provision of breastfeeding support and education: Systematic review and meta-analysis. Maternal & Child Nutrition

The authors declare that there are no conflict of interests.

AG, AM, JM and SR designed the research study. All authors were involved in at least one stage of the research process (i.e., study selection, data extraction, risk of bias assessment, meta-analysis). AG, AM and JM wrote the manuscript. PB and SR helped with re-drafting the manuscript. All authors have read and approved the final manuscript.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

http://orcid.org/0000-0003-3750-2445Joyce Marshall http://orcid.org/0000-0002-2784-1817Alison McFadden http://orcid.org/0000-0002-5164-2025