1. Objectives

To describe and evaluate the effectiveness of different modes of breastfeeding support, provided by professional or lay people, to mothers in the context of varying background breastfeeding initiation rates

2. How studies were identified

The following databases were searched in February 2016:

• Cochrane Pregnancy and Childbirth Group’s Trials Register
• CENTRAL (The Cochrane Library)
• MEDLINE
• EMBASE
• CINAHL

Reference lists were also searched

3. Criteria for including studies in the review

3.1 Study type

Randomized or quasi-randomized controlled trials, including cluster-randomized controlled trials

3.2 Study participants

Healthy pregnant women intending to or considering breastfeeding and healthy women breastfeeding their babies

3.3 Interventions

Support interventions were supplementary to local standard care and could involve verbal assurance and praise; information and response to the mother’s questions; or staff training to improve the support provided to women. Support could be offered by health professionals or lay people, be offered proactively or reactively, be provided in groups or individually, be provided face-to-face or via telephone, and could range from only one contact to ongoing contact over several months

(Interventions taking place solely in the antenatal period were excluded)

3.4 Primary outcomes

Stopping any breastfeeding

• Stopping any breastfeeding before six months postpartum
• Stopping any breastfeeding before four to six weeks postpartum

Stopping exclusive breastfeeding

• Stopping exclusive breastfeeding before six months postpartum
• Stopping exclusive breastfeeding before four to six weeks postpartum

Secondary outcomes were not included in this update of the review

4. Main results

4.1 Included studies

One hundred randomized controlled trials, enrolling over 83,246 mother-infant dyads, were included in this review

• Seventy-three of the 100 trials contributed data to analyses, 15 of which were cluster-randomized, involving a total of 74,656 mother-infant pairs
• In 64 of the 73 studies contributing data, women received the intervention; in eight trials staff received additional training; and in one trial a policy was introduced for providing breastfeeding groups
• In 49 of the 73 studies support was provided by professionals and in most trials support was offered proactively
• Postnatal contacts between support worker and mother ranged from fewer than three to more than nine
• In 57/73 studies there was one-to-one contact between the breastfeeding supporter and breastfeeding mother, two studies used group support, one offered group and individual support, and one offered support to couples
• In 47/73 trials, telephone support was offered, and in four of these studies telephone contact was the sole intervention
• Twenty-four studies recruited women antenatally, of which six trials only included women who intended to breastfeed

4.2 Study settings

• Of the 73 trials contributing data to analyses, five were published prior to 1990, 10 were published between 1990 and 1999, 40 between 2000 and 2011, and 18 were published between 2012 and 2016
• Four studies were conducted in low-income settings (Bangladesh, Burkina Faso and Uganda [2 trials], Democratic Republic of the Congo); four studies were conducted in low-middle-income settings (India, Kenya, Pakistan, Syria); 15 studies were conducted in upper-middle-income settings (Belarus, Brazil [8 trials], China, Iran, Malaysia, Mexico, Turkey, South Africa); 52 studies were conducted in high-income settings (Australia [5 trials], Canada [8 trials], China, Croatia, Denmark, France [2 trials], Italy, the Netherlands [2 trials], Singapore, Sweden [2 trials], the United Kingdom of Great Britain and Northern Ireland [8 trials], the United States of America [20 trials])
• Twenty-eight studies were undertaken in groups of low socioeconomic position within their country
• Studies were carried out in both hospital and community settings; in six hospital-based studies, the control group received Baby Friendly Hospital Initiative standard-of-care
• All country-level income categories were represented in the 24 studies with high rates of breastfeeding initiation (≥80% ever breastfed). However, the four studies from low-/low-middle-income countries all had high rates, and all studies with intermediate (60 to <80%, n=18) or low (<60%, n=11) rates of breastfeeding initiation were conducted in high-income settings

4.3 Study settings

How the data were analysed
Breastfeeding support interventions were compared to standard care. Random effects meta-analysis was used to analyse pooled data. For dichotomous data, risk ratios (RR) and 95% confidence intervals (CI) were calculated. Outcome data where there was attrition of more than 25% were excluded, and intention-to-treat analysis was conducted for primary outcomes, assuming that all women who were lost to follow-up had stopped breastfeeding by given time points. Data from 12 of the 15 cluster-randomized trials were adjusted for clustering. In trials with more than two intervention groups, double-counting of the control group was avoided by dividing it between each relevant intervention group in comparisons. Sensitivity analyses for primary outcomes were conducted by study quality, comparing the effect size in studies with low versus high/unclear risk of bias for allocation concealment. Subgroup analyses were planned as follows:

• By type of supporter: lay person versus professional
• By type of support: face-to-face versus telephone
• By timing of support: postnatal versus antenatal plus postnatal
• By how support was offered: proactively or reactively
• By background breastfeeding rates: low, medium or high
• By intensity of support

Results
Stopping Breastfeeding
Any breastfeeding up to six months postpartum
Overall, in 51 trials including 21,418 women, 53% of those receiving additional breastfeeding support interventions had ceased any breastfeeding by six months postpartum compared with 57% of those who received standard care. Pooled analysis showed a statistically significant 9% reduction in the risk of stopping breastfeeding before six months in the intervention group in comparison to controls (RR 0.91, 95% CI [0.88 to 0.95], I²=55%). In sensitivity analysis of trials at low risk of bias for allocation concealment, the effect remained significant. Funnel plot asymmetry was evident with smaller studies showing a greater treatment effect.

Exclusive breastfeeding up to six months postpartum
Forty-six trials including 18,591 women evaluated the effect of breastfeeding support interventions on exclusive breastfeeding up to six months postpartum, in which 75% of mothers in the intervention group had ceased exclusive breastfeeding by this time in comparison to 83% of controls. In meta-analysis, additional breastfeeding support resulted in a 12% reduction in the risk of stopping exclusive breastfeeding by six months postpartum (RR 0.88, 95% CI [0.82 to 0.92], I²=95%). In sensitivity analysis of trials at low risk of bias for allocation concealment, the effect was reduced but remained significant. Funnel plot asymmetry suggested that smaller studies reported greater treatment effects.

Any breastfeeding and exclusive breastfeeding at four to six weeks postpartum
Women receiving breastfeeding support interventions were less likely to stop any breastfeeding before six weeks postpartum (RR 0.87, 95% CI [0.80 to 0.95], I²=54%; 33 studies/10,776 women), and were also less likely to stop exclusive breastfeeding by this time (RR 0.79, 95% CI [0.71 to 0.89], I²=97%; 32 studies/10,960 women). Sensitivity analyses revealed lower, but still significant, effect sizes for trials at low risk of bias for allocation concealment. Funnel plot asymmetry suggested smaller studies had larger effect sizes and that small studies with null or negative findings may not have been published.

Additional subgroup analysis of breastfeeding cessation
For the outcome of cessation of any breastfeeding up to six months, support from non-health professionals produced a slightly larger effect size (RR 0.85, 95% CI [0.77 to 0.93]) than support from health professionals (RR 0.92, 95% CI [0.89 to 0.96]) or combined support from both (RR 0.97, 95% CI [0.91 to 1.03]), with a borderline statistically significant difference between subgroups (p=0.05). A similar result was observed for cessation of exclusive breastfeeding up to six months postpartum (lay people: RR 0.76, 95% CI [0.65 to 0.87]; professionals: RR 0.92, 95% CI [0.89 to 0.96]; combined support: RR 0.76, 95% CI [0.44 to 1.32]; p=0.02 for subgroup differences). Results were similar for cessation of exclusive breastfeeding by four to six weeks of age (lay people: RR 0.64, 95% CI [0.46 to 0.89]; professionals: RR 0.84, 95% CI [0.75 to 0.95]; combined support: RR 0.94, 95% CI [0.89 to 0.99]; p=0.03 for subgroup differences), but no clear differences were observed for cessation of any breastfeeding at this time (p=0.48 for subgroup differences).

Face-to-face interventions were superior to telephone-only or combined interventions, reducing the risk of stopping exclusive breastfeeding by six months postpartum by 14% (RR 0.86, 95% CI [0.81 to 0.90]), compared to no reduction for telephone only (RR 1.00, 95% CI [0.99 to 1.01]) and a non-significant reduction for combined face-to-face and telephone contact (RR 0.97, 95% CI [0.93 to 1.00]; p<0.01 for subgroup differences). These subgroup differences remained significant for cessation of exclusive breastfeeding by four to six weeks, but not for cessation of any breastfeeding at either time-point.

Cessation of exclusive breastfeeding up to six months was more substantially reduced in settings with high background rates of breastfeeding initiation (RR 0.84, 95% CI [0.80 to 0.90]) compared to areas with intermediate (RR 0.97, 95% CI [0.93 to 1.02]) or low (RR 1.00, 95% CI [0.99 to 1.01]) background initiation rates (p<0.0001 for subgroup differences). Results were similar for cessation of exclusive breastfeeding by four to six weeks (p<0.01 for subgroup differences). Cessation of any breastfeeding by four to six weeks or up to six months did not differ by background breastfeeding rates (both p≥0.26 for subgroup differences).

For cessation of exclusive breastfeeding, four to eight intervention sessions were superior to fewer than four sessions or more than eight sessions (four to six weeks: p=0.05 for subgroup differences; up to six months: p<0.01 for subgroup differences). There was no evidence of any difference by number of intervention sessions for cessation of any breastfeeding (both p≥0.66 for subgroup differences).

No differences were noted between interventions provided antenatally and postnatally versus postnatally alone (all p≥0.15 for subgroup differences). Subgroup analysis by how support was offered (proactive versus reactive) was not performed.

5. Additional author observations*

Overall, the methodological quality of the evidence was mixed, with less than half of all studies (44/100) being at low risk of bias for allocation concealment. Using GRADE criteria, the quality of the evidence for all four primary outcomes was rated as moderate, with downgrading due to statistical heterogeneity. The support interventions, study settings and standards of care included in this review were diverse, and the size of the treatment effects reflected this, varying considerably in different trials and subgroup analyses. Therefore, the average treatment effects presented in the review may not be applicable to all settings.

Breastfeeding support interventions offered to women increase both the duration and exclusivity of breastfeeding. Subgroup analysis suggested that face-to-face support produced a greater treatment effect than telephone support for exclusive breastfeeding, and that high background rates of breastfeeding initiation were associated with an increased effect on exclusive breastfeeding. While subgroup analysis suggested that lay support and a greater number of intervention visits resulted in larger treatment effects, the within-group heterogeneity in these analyses remained high, indicating that these findings should be interpreted with caution.

Future trials should include detailed data on the attributes of the intervention and of standard care, the population group studied, and background breastfeeding initiation rates. Investigations on the effectiveness of different training programmes, cost-effectiveness, and strategies to support women who wish to breastfeed longer than six months are needed.