Breastfeeding Promotion Programs

Published: January 2023 (September 2018 version of page here)

What is the problem?

WHO and UNICEF believe that increasing breastfeeding rates is critical to improving health, nutrition, and mortality outcomes.2 WHO reports that breastfeeding increases the nutrition and immune system of newborns.3 WHO and UNICEF recommend initiation of breastfeeding at birth, exclusive breastfeeding for six months, and continued breastfeeding up to two years of age or beyond.4 While these recommendations apply to all infants worldwide, the risks from inadequate breastfeeding behaviors are likely to be higher in low-resource environments with poor sanitation and worse substitute foods.5 According to UNICEF, around half of newborns in low- and middle-income countries (LMICs) are not breastfed at birth and around half of children younger than six months are not exclusively breastfed.6

What is the program?

In general, breastfeeding promotion programs educate, encourage, and support mothers to breastfeed.7 These programs can be delivered in different ways, including at hospitals by health professionals, door-to-door by community health workers, as part of peer support groups, or through digital tools.8 Programs can vary in timing and intensity.9 Since we believe the costs and effectiveness of different breastfeeding promotion programs can vary substantially by context, this report investigates LMIC breastfeeding promotion programs. We consider both a general direct delivery program (which could include hospital or community-level direct engagement with new or expecting parents), and a general digital program (such as an SMS reminder program).

Does the program have strong evidence of effectiveness?

We have moderate confidence that breastfeeding promotion reduces infant diarrhea and mortality. We conducted a meta-analysis to estimate the program’s effect on diarrhea, and use evidence on the relationship between diarrhea reductions and mortality reductions in water quality interventions to model breastfeeding promotion's mortality effects. We are highly uncertain about the size of this effect, how the program’s effectiveness may vary by context and delivery model, and how long the benefits of breastfeeding promotion persist.

Effects of breastfeeding promotion on breastfeeding

We have moderately high confidence that breastfeeding promotion activities can increase the uptake of breastfeeding, but are uncertain how this could differ across contexts with different baseline breastfeeding rates, and with different programmatic models. Olufunlayo et al. 2019 conducted a meta-analysis on the effect of different breastfeeding interventions on exclusive breastfeeding rates at 6 months, and found that breastfeeding promotion increased exclusive breastfeeding rates by 119%.10 One of the RCTs included in their meta-analysis was a telephone-only intervention, so we re-estimate the meta-analysis excluding this study in order to estimate the effect of direct delivery breastfeeding promotion.11 We estimate that direct delivery breastfeeding promotion increased exclusive breastfeeding rates in the included studies from 18% to 40%.12 Unger et al. 2018 conducted a relatively small randomized controlled trial (RCT) that estimated the effect of an SMS intervention on exclusive breastfeeding rates at 6 months.13 They found that when 1-way SMS communication was used, exclusive breastfeeding rates in the study population increased from 41% to 49%.14

Diarrhea effects

We have moderate confidence that breastfeeding promotion reduces infant diarrhea rates. Based on a moderately deep literature review, we did not identify any recent systematic reviews or meta-analyses evaluating high-quality LMIC evidence of breastfeeding promotion programs’ effects on diarrhea. We identified 10 interventions from 6 publications that linked breastfeeding promotion programs to diarrhea morbidity outcomes, and conducted our own basic meta-analysis on these to estimate diarrhea effects.15 These RCTs took place in LMICs and evaluated a variety of direct delivery models.16 We use this evidence to model diarrhea (and mortality) effects for infants, since it is our understanding that breastfeeding’s benefits are concentrated among this age group.17 Most of these RCTs assessed diarrhea at six months,18 so we make a rough assumption that these findings are representative of post neonatal (28 days to 1 year of age) diarrhea effects. Overall, we estimate that direct delivery breastfeeding promotion reduces diarrhea by around 16% (n=22,471, 95% confidence interval: 1% to 28%).19 There were substantial differences in effects across studies, and we are highly uncertain about what drives differences since studies took place in different contexts and varied along many parameters.20 Breastfeeding rates also improved across all studies, which gives us some additional confidence in the plausibility of breastfeeding promotion reducing diarrhea.21

Mortality effects

We did not identify any high-quality LMIC evidence on breastfeeding promotion’s mortality effects22 , which is the primary benefit we model in our cost-effectiveness analysis. We estimate mortality effects based on the relationship between diarrhea and mortality reductions that we estimate for water quality interventions.23 We spoke with one child health expert, who told us diarrhea is the main mechanism by which breastfeeding reduces mortality.24 Overall, we have moderate confidence breastfeeding reduces mortality, but are uncertain about the magnitude of this effect.

Diarrhea to mortality relationship

We rely on a meta-analysis (Kremer et al. 2022 (working paper)) of 15 RCTs evaluating the effect of water quality interventions on diarrhea and mortality to estimate the relationship between these two effects.25 As noted in our Water Quality Interventions report, we pool the findings of a subset of the included trials to estimate our best guess of the impact of water chlorination on child mortality.26 Here, we use these findings to estimate mortality reductions based on diarrhea reductions for programs which do not have any direct evidence of mortality effects.27

We view Kremer et al. 2022 (working paper) as compelling evidence that (1) water quality interventions reduce mortality and (2) diarrhea and mortality reductions are correlated.28 However, we are uncertain about the specific relationship between these two outcomes across magnitudes.

Application to breastfeeding

Since we did not identify any high-quality LMIC evidence on breastfeeding promotion’s mortality effects, we use Kremer et al. 2022 (working paper) and our estimate of the program’s diarrhea effects to estimate mortality effects. We have some confidence we can apply this estimation technique to breastfeeding, since we believe breastfeeding’s reduction of diarrheal pathogen infection implies a similar mechanism as water quality interventions. This is because breastfeeding may remove (before 6 months) or reduce (6-24 months) the need for complementary feeding, which can sometimes contain harmful water/food-borne pathogens.29

We estimate the size of a plausible mortality effect based on the estimated causes of death for both neonatal and post neonatal infants.30 We assume that mortality from a range of relevant causes is reduced by a percentage equal to the percent reduction in diarrhea morbidity resulting from breastfeeding promotion.31 For additional details, see Water Quality Interventions (and the modeled input to our Water Quality CEA). We are highly uncertain about this estimation technique, and we may investigate it further in future analysis. After adjustments, we estimate that direct delivery of breastfeeding promotion reduces neonatal diarrhea and all-cause mortality by 5.8% and 2.6%, respectively, as well as post neonatal diarrhea and all-cause mortality by 8.4% and 6.8%, respectively.32

We believe it is reasonable to apply this estimation technique to breastfeeding promotion programs, but remain uncertain about whether and how breastfeeding may affect diarrhea and mortality differently than water treatment interventions.33

Additional mortality effects of breastfeeding

We have used our model of the effect of clean water on mortality to develop our best guess of the effect of breastfeeding. However, we are highly uncertain whether we should expect additional mortality effects from breastfeeding through pathways other than those that would affect diarrhea morbidity. Other channels through which breastfeeding could affect mortality may include:

• Non-diarrheal infections. Some research claims that breastfeeding is correlated with several additional health benefits for infants, such as reduced rates of childhood infections in the lungs (e.g. pneumonia) and ears.34 We think it is plausible that breastfeeding could avert a range of nutritional diseases and other infections, and in our model we have assumed that mortality due to a range of relevant causes (including respiratory infections and nutritional deficiencies) could be averted.35 However, we may have underestimated this impact if the magnitude of benefits from breastfeeding for certain outcomes are larger than estimated diarrhea morbidity reductions. We are highly uncertain about the extent to which these various mortality reduction pathways overlap or are already captured in our model.
• Skin-to-skin contact. We think that Kangaroo Mother Care (KMC; an intervention that involves keeping low birth weight infants warm through skin-to-skin contact with their mothers36 ) reduces infant mortality37 , but we do not know whether these mortality benefits would occur with breastfeeding alone.38 We are highly uncertain about how skin-to-skin contact that occurs during breastfeeding would compare with skin-to-skin contact that occurs with KMC.39 However, we do think it is plausible that some skin-to-skin contact through breastfeeding could have some additional benefits not otherwise captured in our model. We have accounted for this with a subjective adjustment (see ‘Additional benefits’ below for more detail).

Internal validity

We view this evidence as having medium-strength internal validity. It is our understanding that the mechanisms by which breastfeeding promotion reduces mortality are highly plausible, and the estimated effect sizes seem plausible to us. We do apply a moderate downward adjustment because of the large variation in the effect sizes found among the evidence we have reviewed, possible risk of response bias, and our heavy reliance on modeling (both in terms of estimating diarrhea morbidity and mapping this to mortality effects). We apply a larger downward adjustment to our digital delivery model because of the limited size of the evidence base we have reviewed. We are very uncertain about the possible uptake in breastfeeding rates caused by less direct programs.

External validity

We have a moderate level of uncertainty about the extent to which findings from trials will generalize to future implementation. Since our findings are based on operational trials across a variety of settings40 , we have some confidence that breastfeeding promotion is not highly context-specific. However, we do make a downward adjustment because we are uncertain what makes breastfeeding promotion more or less effective in different contexts, and generally expect interventions to have relatively smaller effects at scale. In particular, most studies we reviewed occurred in areas with very low baseline breastfeeding rates41 , and we are unsure whether effects would be as large in contexts with higher counterfactual breastfeeding rates. We will take contextual factors into account if we are considering a specific funding opportunity and may revise our external validity estimate accordingly. We apply a larger downward adjustment to our digital delivery model because of the possibility that digital tools are less well targeted to an at-risk population.

Development effects

We did not identify direct evidence of breastfeeding promotion’s effect on long-term productivity and earnings. To estimate effects, we compare the program’s effect on factors correlated with gains in adult consumption to those of other interventions for which we have direct evidence of development effects.42 We estimate development effects of breastfeeding promotion are around 30% of those of seasonal malaria chemoprevention, and account for 12% of breastfeeding promotion’s overall value.43

Additional benefits

We believe breastfeeding promotion may have additional benefits, such as morbidity reduction (including morbidity related to diarrhea, respiratory infections44 and ear infections45 ), birth spacing, and maternal health benefits. We have not investigated these thoroughly because we believe their value is small relative to infant mortality and/or there is a lack of high-quality evidence of these benefits. We have included a rough estimate for additional benefits in our cost-effectiveness analysis and may investigate these further.46

Potential offsetting/negative effects

We believe there are unlikely to be substantial adverse effects associated with this intervention. We did not identify any evidence suggesting breastfeeding promotion generally results in offsetting or negative effects, though there could be higher risk of adverse effects for certain interventions and/or in certain contexts. Some possible risks include:

• A risk of causing mental distress in mothers who feel social pressure to breastfeed but are unable to do so. Some mothers may struggle to breastfeed or face physical constraints including infection and undersupply of milk.
• A risk of HIV transmission to infants via breastfeeding by HIV-positive mothers without access to recommended antiretroviral drug regimens.47
• A risk of accidentally malnourishing a minority of infants (if advice to breastfeed exclusively is given and taken too rigidly) or a risk of compounding malnutrition in mothers in low-resource environments.

We will consider potential context-specific negative effects if we consider a specific funding opportunity.

How cost-effective is the program?

We conducted a preliminary cost-effectiveness analysis. As of December 2022, we estimate that breastfeeding promotion is around the level of cost-effectiveness of programs we expect to direct marginal donations to in digital delivery contexts. We estimate that direct delivery programs would be slightly below our cost-effectiveness threshold. However, we expect our bottom line could change if we get more information (in particular, cost information).48

Note that our cost-effectiveness analyses are simplified models that do not take into account a number of factors. There are limitations to this kind of cost-effectiveness analysis, and we believe that cost-effectiveness estimates such as these should not be taken literally due to the significant uncertainty around them. We provide these estimates (a) for comparative purposes and (b) because working on them helps us ensure that we are thinking through as many of the relevant issues as possible.

This cost-effectiveness analysis is at an early stage, and we think it’s likely that our bottom line cost-effectiveness estimate will change with further review.

Breastfeeding promotion is moderately inexpensive (see below) and leads to a moderate reduction in infant mortality. We have high uncertainty about costs, as well as breastfeeding promotion’s effect on subsequent births and actual effect on all-cause mortality. We modeled both a generic direct delivery program (e.g. support at hospitals by health professionals, door-to-door by community health workers, or as part of a peer support group), and an SMS program (where we assumed that texts were one-way and targeted at women who may be interested in the program). Different delivery models could have substantially different costs and program uptake.

A sketch of the cost-effectiveness model is below:

• Baseline mortality. We use average all-cause mortality rates for low socio-demographic index countries, as defined by the Institute for Health Metrics and Evaluation: 3% for neonates and 2% for post neonatal infants.49
• Mortality effect. We estimate that breastfeeding promotion may plausibly reduce neonatal mortality by 3% and post neonatal mortality by 7% when the program is directly delivered to parents. We estimate that the effects are smaller (a 1% reduction in neonatal mortality and a 2% reduction in post neonatal mortality) when breastfeeding is promoted via SMS. This difference is driven by differences in breastfeeding uptake.50
• Additional benefits. We estimate development effects account for 12% of the program’s overall value and small adjustments for excluded effects account for an additional 11%.51
• Costs. We estimate direct delivery breastfeeding promotion costs around $28 per mother/child pair treated.52 We estimate that breastfeeding promotion via SMS costs around $3 per mother/child pair treated.53 We also include a multiplier to account for mothers increasing breastfeeding for roughly one additional subsequent child, which reduces costs to around $14 per child treated in a direct delivery program or $1.51 per child treated in an SMS program.54
• Cost-effectiveness. Using an estimate of the overall cost to all contributors of $14 per mother/child pair treated, our best guess is that a direct delivery breastfeeding promotion program could avert a death (including neonatal and post neonatal deaths) for an average of $5,943.55 Using an estimate of the overall cost to all contributors of $1.51 per mother/child pair treated, our best guess is that digital breastfeeding promotion programs could avert a death (including neonatal and post neonatal deaths) for an average of $2,066.56

We have high uncertainty about costs, as well as breastfeeding promotion’s effect on subsequent children, the actual effect on all-cause mortality, and the duration of benefits.

• Costs. We estimate costs for a direct delivery program based on estimates from one study.57 We have high uncertainty about this estimate and believe costs likely vary substantially based on a program’s context, scale, intensity, and delivery model. We estimate costs for a digital delivery program based largely on a similar mobile-based program we have modeled58 , but are also highly uncertain about how applicable this may be to programs in other contexts. We expect to be able to refine our estimates if we investigate a specific giving opportunity.
• All-cause mortality effects. We estimate mortality based on breastfeeding promotion’s diarrhea effects. We are highly uncertain about the specific effect breastfeeding promotion has on diarrhea, since effects vary substantially across the interventions we evaluated.59 We are additionally highly uncertain about how breastfeeding may affect diarrhea and mortality across different age ranges. While we have limited confidence in this estimate, we believe it is a reasonable approach at this stage of our investigation. We expect to continue to refine this estimation as we further investigate interventions that reduce diarrhea.
• Effect on subsequent children. Based on a light literature review, we did not identify any evidence on breastfeeding promotion’s effect on subsequent births.60 We spoke with one child health expert who believes there is such an effect.61 Our current multiplier effect is based on a set of assumptions about which we are highly uncertain, and which may also vary based on program and context (e.g., targeting first-time mothers, context-specific fertility rates). This multiplier effect may overestimate effects if fadeout effects are larger, or if the intervention reaches mothers who have already received the program and are less likely to change behavior due to receiving breastfeeding promotion an additional time. On the other hand, effects may be larger for subsequent children due to positive feedback (e.g., healthy baby) or reaching a program intensity threshold that causes behavior change only after repeat interventions. It is also possible there is spillover information spread between mothers as a result of the program, which could increase this multiplier effect further.62 Given the lack of evidence and unclear direction of variables potentially influencing this effect, we believe our current multiplier is a reasonable best guess.
• Duration of benefits. We are highly uncertain about the duration of benefits that are likely to occur due to increased breastfeeding in early infancy. We made a rough assumption that studies that measured effects at 6-months old were representative of the post neonatal (28 days to 1 year of age) period. Mortality effects could be less than expected if women stop breastfeeding while children are between 6 and 12 months old. Conversely, if women continue to breastfeed beyond children being 1 year of age, it is possible that mortality effects could be larger than currently modeled. We do not have a strong prior on whether we are over- or underestimating effects at this stage.

Does the program appear to have room for more funding?

We have not yet estimated this program’s room for more funding. Based on breastfeeding promotion’s large target population and current breastfeeding practices, there may be substantial room for more funding, though we are highly uncertain about this and have not investigated the degree to which financial resources are a limiting factor in breastfeeding promotion’s scale-up.63 There may be opportunities to support government-led programs via technical assistance, as well as directly fund programs implemented by NGOs. We plan to continue to investigate other potential funding opportunities.

Key questions for further investigation

As we deepen our investigation of this program, we will focus our research on the following areas:

• In what contexts is breastfeeding promotion likely to be most effective?
• In what contexts and under what circumstances are specific breastfeeding promotion delivery models likely to be most effective?
• What are the costs of different breastfeeding promotion delivery models?
• What is the effect of breastfeeding promotion on subsequent children?
• Is the link between diarrhea morbidity to all-cause mortality likely to be similar between breastfeeding and water quality interventions (in particular because of the age distribution of children involved)?
• Is technical assistance to support government-led breastfeeding promotion programs cost-effective?
• What organizations and governments have room for more funding to scale breastfeeding promotion programs?

Our process

We conducted a moderately deep literature review to assess evidence on breastfeeding promotion’s effects on breastfeeding, diarrhea, and mortality. We used individual high-quality LMIC studies evaluating diarrhea to conduct our own meta-analysis. We relied on the broader evidence base to qualitatively sense-check our findings.

Sources

Document	Source
Bashour et al. 2008	Source
Bhandari et al. 2003	Source
Chan et al. 2016	Source (archive)
Chola et al. 2011	Source (archive)
Deeks, Higgins, and Altman, Chapter 10: Analysing data and undertaking meta-analyses, Cochrane Handbook for Systematic Reviews of Interventions version 6.3, 2022	Source (archive)
GiveWell, "IRD Global – Mobile conditional cash transfers for immunizations (October 2021)"	Source
GiveWell, "Kangaroo Mother Care," 2021	Source
GiveWell, "Water quality interventions," 2022	Source
GiveWell, Breastfeeding promotion CEA, 2022	Source
GiveWell, Breastfeeding promotion mortality plausibility modeling	Source
GiveWell, Development effects estimation method	Source
GiveWell's non-verbatim summary of a conversation with Dr. Thorkild Tylleskär, May 21, 2021	Source
Horta and Victora 2013	Source (archive)
Institute for Health Metrics and Evaluation, "Socio-demographic index (SDI)"	Source (archive)
Jakobsen et al. 2007	Source
Kramer et al. 2001	Source
Kremer et al. 2022 (working paper)	Source (archive)
McFadden et al. 2017	Source
Morrow et al. 1999	Source
Olufunlayo et al. 2019	Source
Qian et al. 2021	Source (archive)
Sutter Health, "How often to breastfeed"	Source (archive)
The Centers for Disease Control and Prevention, "Maternal or infant illnesses or conditions: HIV"	Source (archive)
The World Bank, "Birth rate, crude (per 1000 people) - Low & middle income"	Source (archive)
The World Bank, "Fertility rate, total (births per woman) – Low income"	Source (archive)
The World Bank, "Population, total - Low & middle income"	Source (archive)
The World Health Organization and United Nations Children's Fund, Implementation Guidance: Baby-Friendly Hospital Initiative, 2018	Source (archive)
The World Health Organization, "Drinking-water," 2022	Source (archive)
The World Health Organization, "Infant and young child feeding fact sheet," 2021	Source (archive)
The World Health Organization, "WHO's first ever global estimates of foodborne diseases find children under 5 account for almost one third of deaths," 2015	Source (archive)
The World Health Organization, Infant and Young Child Feeding: Model Chapter for Textbooks for Medical Students and Allied Health Professionals, 2009	Source (archive)
The World Health Organization, Kangaroo mother care: A practical guide, 2003	Source (archive)
Tylleskär et al. 2011	Source
Unger et al. 2018	Source
UNICEF, Global database on infant and young child feeding, “Early initiation of breastfeeding (birth)” spreadsheet, 2021	Source (archive)
UNICEF, Global database on infant and young child feeding, “Exclusive breastfeeding ( 6 months)” spreadsheet, 2021	Source (archive)
Victora et al. 2016	Source
WHO Immediate KMC Study Group 2021	Source
Yotebieng et al. 2015	Source (archive)

• 1

  As of the writing of this page (November 2022), GiveWell expects that the cost-effectiveness of its marginal donations is approximately ten times as cost-effective as unconditional cash transfer programs, such as that of GiveDirectly.

• 2

  "Breastfeeding is critical for achieving global goals on nutrition, health and survival, economic growth and environmental sustainability. . . . Inadequate breastfeeding practices significantly impair the health, development and survival of infants, children and mothers. Improving these practices could save over 820 000 lives a year." The World Health Organization and United Nations Children's Fund, Implementation Guidance: Baby-Friendly Hospital Initiative, 2018, p. 8.

• 3

  “Breast milk contains all the nutrients that an infant needs in the first 6 months of life. . . . Breast milk also contains bioactive factors that augment the infant's immature immune system.” The World Health Organization, Infant and Young Child Feeding: Model Chapter for Textbooks for Medical Students and Allied Health Professionals, 2009.

• 4

  “WHO and UNICEF recommend: early initiation of breastfeeding within 1 hour of birth; exclusive breastfeeding for the first 6 months of life; and introduction of nutritionally-adequate and safe complementary (solid) foods at 6 months together with continued breastfeeding up to 2 years of age or beyond.” The World Health Organization, "Infant and young child feeding fact sheet," 2021.

• 5

  "Several mechanisms for a possible protective effect of breastfeeding against gastrointestinal infections have been proposed, including the presence in breastmilk of substances with antimicrobial or immunological properties, avoidance of contamination (as in non-human milk or baby bottles), and the general nutritional status of breastfed infants. . . .

  [N]on-breastfed infants are more exposed to pathogens that may cause diarrhea than breastfed subjects. Many studies attest to the presence of pathogens in foods offered to infants. For example, in The Gambia, Rowland et al observed that weaning foods traditionally given to children were contaminated with microorganisms that could cause gastrointestinal infections. Another study from Chile showed that most feeding bottles harbored large numbers of pathogens that could cause gastrointestinal infection.

  Last, it has been proposed that in low-income settings optimal breastfeeding practices can prevent undernutrition associated with repeated infections and with the use of over-diluted breastmilk substitutes. Good nutrition is essential for non-specific immunity that contributes to fighting infections in general." Horta and Victora 2013, p. 12.

• 6
  • Initiation at birth: UNICEF Global database on infant and young child feeding, “Early initiation of breastfeeding (birth)” spreadsheet, “Latest Regional Global” tab, “World Bank (FY22) income grouping aggregates” section, average of "Low Income," "Lower Middle Income," and "Upper Middle Income" rows. UNICEF, Global database on infant and young child feeding, “Early initiation of breastfeeding (birth)” spreadsheet, 2021.
    Average: (59 + 49 + 39)/3 = 49
  • Breastfeeding at six months: UNICEF, Global database on infant and young child feeding, “Exclusive breastfeeding (< 6 months)” spreadsheet, “Latest Regional Global” tab, “World Bank (FY22) income grouping aggregates” section, average of "Low Income," "Lower Middle Income" and "Upper Middle Income" rows. UNICEF, Global database on infant and young child feeding, “Exclusive breastfeeding ( 6 months)” spreadsheet, 2021.
    Average: (52 + 50 + 26)/3 = 42.67

• 7

  “‘Support’ is complex and can include several elements such as emotional and esteem-building support (including reassurance and praise), practical help, informational support (including the opportunity to discuss and respond to women’s questions) and social support (including signposting women to support groups and networks).” McFadden et al. 2017, p. 7.

• 8
  • “It can be offered in a range of ways, by health professionals or lay people, trained or untrained, in hospital and community settings. It can be offered to groups of women or one-to-one, it can involve mother-to-mother support, and it can include family members (typically fathers or grandmothers) and wider communities.” McFadden et al. 2017, p. 7.
  • “Many factors have been identified as having an impact on breastfeeding outcomes, and a key to solving the problem of the low breastfeeding rate is to improve awareness among pregnant women and mothers, as well as perform regular follow-ups. Face-to-face interventions require high levels of cooperation in postpartum women, and it is easy for women to be lost to follow-up. One proposed solution is mobile health (mHealth), which could provide medical assistance with the help of electronic mobile devices. Compared to face-to-face medical assistance, mHealth is cheaper and can have improved compliance.”Qian et al. 2021

• 9

  “Support can be offered proactively by contacting women directly, or reactively, by waiting for women to get in touch. It can be provided face-to-face, by telephone or through social media. It can involve only one contact or regular, ongoing contact over several months.”
  McFadden et al. 2017, p. 7.

• 10

  "Exclusive breastfeeding (EBF) rates until 6 months in most low and middle income countries (LMICs) are well below the 90% World Health Organization benchmark. This systematic review sought to provide evidence on effectiveness of various interventions on EBF until 6 months in LMICs, compared with standard care. . . . At 6 months, intervention group infants were more likely to be exclusively breastfed than controls (RR = 2.19, 95% CI [1.73, 2.77]; I2 78.4%; 25 randomized controlled trials).” Olufunlayo et al. 2019, p. 1.

• 11

  For this analysis, we used results only from RCTs that assessed the primary outcome of exclusive breastfeeding rates at 6 months. See the list of included RCTs in Olufunlayo et al. 2019, Figure 2, p. 13. Out of this list of 25 RCTs, only one was a telephone-only intervention. The reference numbers for telephone-only interventions were 32, 43, and 56 (see Table 1, section "Mode of delivery of intervention," row "Telephone (voice/SMS)"). According to Appendix B, the telephone-only interventions were Jiang et al. 2014 (see p. 20, reference number 32), Maslowsky et al. 2016 (see p. 20, reference number 43), and Tahir and Al-Sadat 2013 (see p. 21, reference number 56). Out of these three studies, only Tahir and Al-Sadat 2013 is included on the list of RCTs assessing rates of exclusive breastfeeding at 6-months (see Figure 2, p. 13). Olufunlayo et al. 2019.

• 12

  See Breastfeeding promotion CEA, “Uptake of breastfeeding” sheet, for our calculations of the effect of these interventions on rates of exclusive breastfeeding at 6 months. Using the random effects model, RR was calculated as 2.26, with a 95% confidence interval of [1.81, 2.81].

• 13

  "Population–Three hundred women attending antenatal care were randomised, 100 to each arm, and followed for 24 weeks postpartum. Pregnant women 14 years old with access to a phone and able to read SMS were eligible for participation.

  Methods–Women were randomised (1:1:1) to receive 1-way SMS versus 2-way SMS with a nurse versus control. Weekly SMS content was tailored for maternal characteristics and pregnancy or postpartum timing.

  Main Outcome Measures–Facility delivery, EBF and contraceptive use were compared separately between each intervention arm and the control arm by Kaplan-Meier analysis and X2 tests using intent-to-treat analysis." Unger et al. 2018, p. 2.

• 14

  "The probability of EBF to 24 weeks postpartum was higher in both intervention groups than in the control, but only statistically significant in the 2-way messaging group [0•49 in 1-way, 0•62 in 2-way, and 0•41 in control, (p=0•30 and 0•005 for 1-way and 2-way vs.control, respectively)] (Table 2)." Unger et al. 2018, p. 6.

• 15

  We included all studies that we identified that ran a breastfeeding promotion program in a LMIC and reported diarrhea rates. See Breastfeeding promotion CEA, “Meta-analysis” sheet.

  The six publications were Tylleskär et al. 2011 (including RCTs in Burkina Faso, Uganda, and South Africa), Kramer et al. 2001, Yotebieng et al. 2015 (including two different intervention arms), Bhandari et al. 2003, Morrow et al. 1999, and Bashour et al. 2008 (including two different intervention arms). See Breastfeeding promotion CEA, “Evidence from meta-analysis” sheet, for further details on the included studies.

• 16

  See Breastfeeding promotion CEA, “Evidence from meta-analysis” sheet, for study characteristics.

• 17

  We also include a rough guess of benefits for those aged 1-2, which we believe are small compared to benefits for infants. See Breastfeeding promotion CEA, “Adjustments for excluded effects” sheet, "All-cause mortality for children aged 1+" row.

• 18
  • "The primary outcomes were prevalence of EBF and diarrhoea reported by mothers for infants aged 12 weeks and 24 weeks." Tylleskär et al. 2011 .
  • "Main Outcome Measures: Duration of any breastfeeding, prevalence of predominant and exclusive breastfeeding at 3 and 6 months of life and occurrence of 1 or more episodes of gastrointestinal tract infection, 2 or more episodes of respiratory tract infection, and atopic eczema during the first 12 months of life, compared between the intervention and control groups." Kramer et al. 2001, p. 413.
  • "We also assessed two secondary outcomes: the prevalence of infants with reported diarrhoea and the prevalence with reported respiratory illness (fever with cough) between 10 and 14 weeks and between 18 and 24 weeks post partum" Yotebieng et al. 2015, p. 549.
  • "Feeding at age 3 months, and anthropometry and of diarrhoea prevalence at age 3 months and 6 months were assessed." Bhandari et al. 2003.
  • "Outcomes were assessed at 4 months postpartum, through visits carried out by two female trained social scientists who administered a pretested semistructured questionnaire. Primary outcomes included: (1) maternal postpartum morbidities; (2) postnatal care uptake; (3) contraceptive uptake and type; (4) infant morbidities; (5) infant immunization according to the national schedule at 3 months; and (6) infant feeding, namely exclusive breastfeeding during the first 4 months of life." Bashour et al. 2008, p. 117.
  • "Secondary outcomes were duration of breastfeeding, the proportion of infants who had an episode of diarrhoea in the first 3 months (cumulative incidence), and maternal satisfaction with counselling." Morrow et al. 1999, p. 1227.

• 19

  Using the random effects model, RR was calculated as 0.84, with a 95% confidence interval of [0.72; 0.99]. See Breastfeeding promotion CEA, “Meta-analysis” sheet.

  For calculation of the total sample size, see here.

• 20
  • Meta-analysis I2 = 0.72, which the Cochrane Handbook roughly considers “substantial heterogeneity.” For I2 value, see Breastfeeding promotion CEA, “Meta-analysis” sheet. "Thresholds for the interpretation of the I2 statistic can be misleading, since the importance of inconsistency depends on several factors. A rough guide to interpretation in the context of meta-analyses of randomized trials is as follows:
    • 0% to 40%: might not be important;
    • 30% to 60%: may represent moderate heterogeneity*;
    • 50% to 90%: may represent substantial heterogeneity*;
    • 75% to 100%: considerable heterogeneity*." Deeks, Higgins, and Altman, Chapter 10: Analysing data and undertaking meta-analyses, Cochrane Handbook for Systematic Reviews of Interventions version 6.3, 2022.
  • See Breastfeeding promotion CEA, “Evidence from meta-analysis” sheet, for differences in study characteristics.
  • Our uncertainty is similar to that reported in McFadden et al. 2017, a Cochrane Review that evaluated breastfeeding effects for 73 RCTs, mostly from high-income countries:
    • “We investigated substantial heterogeneity for all four outcomes with subgroup analyses for the following covariates: who delivered care, type of support, timing of support, background breastfeeding rate and number of postnatal contacts. Covariates were not able to explain heterogeneity in general. Though the interaction tests were significant for some analyses, we advise caution in the interpretation of results for subgroups due to the heterogeneity. . . . because within-group heterogeneity remained high for all of these analyses, we advise caution when making specific conclusions based on subgroup results.” McFadden et al. 2017, p. 2.
    • "The 73 trials that contributed to the analyses were from 29 countries and involved 74,656 women. Some 62% of the women were from high-income countries[.]” McFadden et al. 2017, pp. 2-3.

• 21

  See Breastfeeding promotion CEA, “Evidence from meta-analysis” sheet, "Effect on breastfeeding rates" column.

• 22

  Jakobsen et al. 2007 reported on an RCT that did not find a mortality effect from a program that encouraged exclusive breastfeeding use: “Introduction of both water and weaning food was significantly delayed in the intervention group. However we found no beneficial health effects of the intervention; there was no reduction in mortality in the intervention group compared with the control group (mortality rate ratio: 1.86 (0.79–4.39)), weight at 4–6 months of age was significantly lower in the intervention group (7.10 kg vs. 7.25 kg; Wilcoxon two-sample test: p = 0.03). There was no difference in diarrhoea morbidity and hospitalization rates.” Jakobsen et al. 2007, p. 68. However, the rates of exclusive breastfeeding in both the control and treatment arms were very low, with 99% (561/566; control group) and 98% (560/569; intervention group) of infants having been introduced to water by the time they were 6 months old (see Table 1, p. 71).

• 23

  See Breastfeeding promotion CEA, "Mortality plausibility modeling" sheet.

• 24

  “Reduction in enteric infections as the primary mechanism by which breastfeeding promotion lowers infant mortality. This is due to a reduction in consumption of contaminated water, which has many protective effects.” GiveWell's non-verbatim summary of a conversation with Dr. Thorkild Tylleskär, May 21, 2021.

• 25

  “We identified 52 RCTs of water quality interventions and included 15 RCTs that had data on child mortality in this meta-analysis.” Kremer et al. 2022 (working paper), p. 2.

• 26

  “To limit these concerns and generate an estimate that is most applicable to the specific interventions we are evaluating, we developed an alternative meta-analysis method in consultation with our external reviewers. We pool the findings of a subset of the trials identified by Kremer et al. 2022 (working paper) that have the following characteristics:

  • The water treatment method is chlorination, without additional treatments like flocculation or filtration. This excludes water quality interventions that were less similar to those we are evaluating.
  • Follow-up length of one year or greater. This tends to exclude small trials that are more susceptible to publication bias.” GiveWell, "Water quality interventions," 2022.

• 27

  See Breastfeeding promotion CEA, “Mortality plausibility modeling” sheet.

• 28

  "The resulting estimate suggests that chlorination interventions reduce all-cause mortality in children under five by approximately 14% in low-income settings (95% confidence interval, 32% reduction to 10% increase). . . . Stronger evidence suggests that chlorination reduces the risk of experiencing diarrhea in children under five. Since diarrhea is a major cause of mortality in this age group, we are fairly confident that chlorination reduces all-cause mortality to some degree." See our Water Quality Interventions report.

• 29

  For example:

  • "Contaminated water and poor sanitation are linked to transmission of diseases such as cholera, diarrhoea, dysentery, hepatitis A, typhoid, and polio. . . . Some 829 000 people are estimated to die every year from diarrhoea as a result of unsafe drinking-water, sanitation and hand hygiene." The World Health Organization, "Drinking-water," 2022
  • "Children are at particular risk of foodborne diarrhoeal diseases, with 220 million falling ill and 96 000 dying every year. . . . The risk of foodborne diseases is most severe in low- and middle-income countries, linked to preparing food with unsafe water; poor hygiene and inadequate conditions in food production and storage; lower levels of literacy and education; and insufficient food safety legislation or implementation of such legislation." The World Health Organization, "WHO's first ever global estimates of foodborne diseases find children under 5 account for almost one third of deaths," 2015.

• 30

  See Breastfeeding promotion CEA, “Mortality plausibility modeling” sheet.

• 31

  See the column "Plausible reduction in all-cause mortality from BP intervention" here.

• 32

  See Breastfeeding promotion CEA, “CEA” sheet.

• 33

  It could be the case that breastfeeding may have additional or larger effects than water quality interventions through improved nutrition and immune systems (such as by not transferring chlorine-resistant microorganisms or generally increasing nutrition). To investigate this, we assessed whether breastfeeding promotion may have relatively larger effects on more severe diarrhea cases. Two of the publications we rely on for our estimate of diarrhea effects report both less severe cases (incidences) and more severe cases (sought treatment / hospitalized). See Bhandari et al. 2003, Table 4, p. 1422. and Kramer et al. 2001, Table 4, p. 418. These two studies found conflicting results (see here). Given this and the limited evidence base, we view this analysis as inconclusive.

• 34
  • "In terms of child morbidity, overwhelming evidence exists from 66 different analyses, mostly from LMICs and including three randomised controlled trials, that breastfeeding protects against diarrhoea and respiratory infections. About half of all diarrhoea episodes and a third of respiratory infections would be avoided by breastfeeding.” Victora et al. 2016, p. 479.
  • "Our reviews suggest important protection against otitis media [ear infections] in children younger than 2 years of age, mostly from high-income settings, but inconclusive findings for older children.” Victora et al. 2016, p. 479.

• 35

  See Breastfeeding promotion CEA, “Mortality plausibility modeling” sheet. For more information on how we assess the plausibility of breastfeeding promotion programs reducing mortality, see this document.

• 36

  “'Kangaroo mother care,' defined as both continuous skin-to-skin contact of the infant with the chest of the mother (or another caregiver when not possible with the mother) and feeding exclusively with breast milk, is among the most effective interventions for preventing death in infants with low birth weight. World Health Organization (WHO) guidelines currently recommend initiation of short, intermittent sessions of kangaroo mother care when the infant’s condition begins to stabilize and continuous kangaroo mother care when the infant’s condition has stabilized.” WHO Immediate KMC Study Group 2021.

• 37

  See our report on Kangaroo Mother Care for more information.

• 38

  In addition to skin-to-skin contact, KMC can also include exclusive breastfeeding and relaxing the criteria for discharge from the facility in order to reduce length of hospitalization, though these components of KMC are not implemented consistently in studies on KMC: “We screened 1035 articles and reports; 299 contained data on KMC and neonatal outcomes or qualitative information on KMC implementation. Eighty–eight of the studies (29%) did not define KMC. Two hundred and eleven studies (71%) included skin–to–skin contact (SSC) in their KMC definition, 49 (16%) included exclusive or nearly exclusive breastfeeding, 22 (7%) included early discharge criteria, and 36 (12%) included follow–up after discharge.”
  Chan et al. 2016, p. 1. To the extent that some benefits from KMC come directly from an increase in exclusive breastfeeding, this should not be additive to the mortality benefits already considered.

• 39

  We are uncertain about the amount of skin-to-skin contact that occurs during breastfeeding, as compared to KMC, and therefore whether we can plausibly expect large mortality reductions. One hypothesized mechanism driving KMC’s impact on neonatal outcomes is that skin-to-skin contact with the mother regulates the infant’s body temperature, reducing hypothermia (“Studies carried out in low-income countries show that prolonged skin-to-skin contact between the mother and her preterm/LBW infant, as in KMC, provides effective thermal control and may be associated with a reduced risk of hypothermia.” The World Health Organization, Kangaroo mother care: A practical guide, 2003, p. 8.). We are highly uncertain whether temperature regulation could be expected from breastfeeding alone, where skin-to-skin contact is typically of a shorter duration:

  • “Skin-to-skin contact should start gradually, with a smooth transition from conventional care to continuous KMC. Sessions that last less than 60 minutes should, however, be avoided because frequent changes are too stressful for the baby. The length of skin-to-skin contacts gradually increases to become as continuous as possible, day and night, interrupted only for changing diapers, especially where no other means of thermal control are available.” The World Health Organization, Kangaroo mother care: A practical guide, 2003, p. 24.
  • “During the newborn period, most breastfeeding sessions take 20 to 45 minutes.” Sutter Health, "How often to breastfeed"

• 40

  See Breastfeeding promotion CEA, “Evidence from meta-analysis” sheet, for settings of the studies included in our meta-analysis.

• 41

  The average control group exclusive breastfeeding rate across the studies we reviewed was 11%. See our calculations here.

  The average rate of exclusive breastfeeding in low- and middle-income countries is estimated by WHO to be around 43%. See Breastfeeding at six months: UNICEF, Global database on infant and young child feeding, “Exclusive breastfeeding (< 6 months)” spreadsheet, “Latest Regional Global” tab, “World Bank (FY22) income grouping aggregates” section, average of "Low Income," "Lower Middle Income," and "Upper Middle Income" rows. @UNICEF, Global database on infant and young child feeding, “Exclusive breastfeeding ( 6 months)” spreadsheet, 2021@.
  Average: (52 + 50 + 26)/3 = 42.67

• 42

  The factors we believe may correlate with gains in adult consumption are direct evidence for adult consumption effects, cognitive ability, weight gain, all-cause mortality, morbidity, anemia, and plausibility. See Development effects estimation method for additional details.

• 43

  See Breastfeeding promotion CEA, “Development effects” sheet.

• 44

  "In terms of child morbidity, overwhelming evidence exists from 66 different analyses, mostly from LMICs and including three randomised controlled trials, that breastfeeding protects against diarrhoea and respiratory infections. About half of all diarrhoea episodes and a third of respiratory infections would be avoided by breastfeeding.” Victora et al. 2016, p. 479.

• 45

  "Our reviews suggest important protection against otitis media [ear infections] in children younger than 2 years of age, mostly from high-income settings, but inconclusive findings for older children.” Victora et al. 2016, p. 479.

• 46

  See Breastfeeding promotion CEA, “Adjustments for excluded effects” sheet.

• 47

  “In resource-limited settings, such as some parts of Africa, the World Health Organization (WHO) recommends that HIV-infected mothers breastfeed exclusively for the first 6 months of life and continue breastfeeding for at least 12 months, with the addition of complementary foods. These mothers should be given ART to reduce the risk of transmission through breastfeeding[.]” The Centers for Disease Control and Prevention, "Maternal or infant illnesses or conditions: HIV"

• 48
  • For an example of the cost-effectiveness of our recommendations, see this page. As of December 2022, we estimate that the cost-effectiveness of opportunities we direct funding to is 10 times as cost-effective as unconditional cash transfers.
  • Our cost-effectiveness analysis estimates that this program ranges from 5.8 to 16.7 times as cost-effective as unconditional cash transfers, depending on the delivery model. See our Breastfeeding promotion CEA, “CEA” sheet, "Cost-effectiveness of breastfeeding promotion in multiples of cash" row.

• 49

  For additional information on the socio-demographic index, see Institute for Health Metrics and Evaluation, "Socio-demographic index (SDI)"

• 50

  We assume that the breastfeeding rates increase by a smaller amount when promoted through digital tools as compared with in-person support.

  • Olufunlayo et al. 2019 estimated the effect of different breastfeeding interventions on exclusive breastfeeding rates at 6 months, and found that breastfeeding promotion increased exclusive breastfeeding rates by 119%. Out of the 25 RCTs included in this analysis, one was a telephone-only intervention. We exclude this study from our calculations, and estimate that direct delivery breastfeeding promotion programs increased exclusive breastfeeding rates across 24 studies from 18% to 40%. See calculations here.
  • Unger et al. 2018 conducted a relatively small RCT that estimated the effect of an SMS intervention on exclusive breastfeeding rates at 6 months. "Three hundred women attending antenatal care were randomised, 100 to each arm, and followed for 24 weeks postpartum. . . . Women were randomised (1:1:1) to receive 1-way SMS versus 2-way SMS with a nurse versus control." Unger et al. 2018, p. 2. This study found that when 1-way SMS communication was used, exclusive breastfeeding rates increased from 41% to 49%. See here.

  This implies that direct delivery (i.e. in-person) breastfeeding promotion reduced the share of children not being breastfed by 27%, while one-way SMS messages reduced the share of children not being breastfed by 14%. As such, we assume that SMS promotion of breastfeeding only reduces diarrhea morbidity by around 50% (13.56 divided by 27.36) as much as direct delivery breastfeeding promotion.

• 51

  See here.

• 52

  This is based on estimates from Chola et al. 2011, which reported on costs of the Uganda arm of the PROMISE EBF trial.

  • "Promise EBF was a multi-centre community randomised trial conducted in four sub-Saharan African countries, namely Burkina Faso, South Africa, Uganda and Zambia. This paper presents the annual costs of the PROMISE EBF intervention in Uganda, and provides estimates of the resources required to fund the scale up to district level." Chola et al. 2011, p. 2.
  • "The cost per mother counselled was US$139 and the cost per visit was US$26.” Chola et al. 2011, p. 1.
  • However, Chola et al. indicated that costs could be lower when implemented in an operational (non-vertical) setting: “We estimated that implementing an alternative package modelled on routine public health sector programmes can potentially reduce costs by over 60%.” Chola et al. 2011, p. 1.

  Our impression is that this trial involved a more intensive form of breastfeeding promotion, so we reduced these estimated costs by half. $139 x (1 - 60%) = $55.6. $55.6 divided by 2 = $27.8. This gives us an estimated cost per mother/child pair of $28. We are highly uncertain about this estimate.

• 53

  This is based on a guess that direct SMS costs would be around $0.28, and operational and other costs would be around $2.73. $0.28 + $2.73 = $3.01.

  • SMS costs are based on an assumption that texts cost $0.01 per SMS, and would be sent weekly from 1 month before birth to 6 months after birth. ($0.01 per SMS x 4 messages per month x 7 months = $0.28)
  • We have proxied operational costs by using our estimate of IRD's total (non-incentive) costs per each eligible infant for a mobile conditional cash transfer program. See here.

• 54

  We spoke with one child health expert who believes it is reasonable to assume that skills learned from breastfeeding support programs would transfer at least in part to a woman's subsequent births. "Based on the assumption that the intervention will affect a new mother's decision to breastfeed not only that child but any subsequent children, GiveWell models that the intervention will result in one more child immediately benefiting and one more child indirectly benefiting on average. Dr. Tylleskär agrees that there are carryover effects, though is uncertain about the size. The effect will be larger in populations with higher birth rates, and the ~25% fadeout effect GiveWell estimates is likely conservative." GiveWell's non-verbatim summary of a conversation with Dr. Thorkild Tylleskär, May 21, 2021. This parameter captures such an effect.

  A multiplier of two assumes one additional child benefits from the program. If this program reaches mothers halfway through their childbearing years, on average, this intervention could plausibly affect half of all children born. In low-income countries, the average fertility rate is 4.5 (see The World Bank, "Fertility rate, total (births per woman) – Low income" year 2020), which implies this program could affect, on average, 2.25 births. We did not identify any evidence on the effect of breastfeeding promotion on subsequent births, but believe it is likely there may be some fadeout effect for subsequent births. As a rough guess, we assume the program has 75% of its original effect on subsequent births, which implies a 2x overall multiplier. A multiplier of two halves the costs per mother/child pair treated from $27.8 to $13.9 (for direct delivery programs) and from $3.01 to $1.51 (for SMS delivery programs).

• 55

  See here for full calculations.

• 56

  See here for full calculations.

• 57

  We based our estimate of costs for a direct delivery program on the findings of Chola et al. 2011.

• 58

  We based our estimate of costs for a digital delivery program on our cost estimates for IRD's mobile conditional cash transfers for immunizations program.

• 59

  See here.

• 60

  We also did not identify any evidence of the effects of maternal health behavior change interventions on subsequent births, which may have qualitatively informed our understanding of potential effects.

• 61

  "Based on the assumption that the intervention will affect a new mother's decision to breastfeed not only that child but any subsequent children, GiveWell models that the intervention will result in one more child immediately benefiting and one more child indirectly benefiting on average. Dr. Tylleskär agrees that there are carryover effects, though is uncertain about the size. The effect will be larger in populations with higher birth rates, and the ~25% fadeout effect GiveWell estimates is likely conservative." GiveWell's non-verbatim summary of a conversation with Dr. Thorkild Tylleskär, May 21, 2021.

• 62

  We did not identify any evidence on this, and the degree of potential spillover is likely dependent on the program's coverage strategy.

• 63

  A rough estimation of theoretical total funding suggests ~$600 million could be devoted annually to breastfeeding promotion.

  • Total population in LMICs in 2021: ~6.6 billion. The World Bank, "Population, total - Low & middle income" year 2021.
  • Birth rate in LMICs: 19/1000. The World Bank, "Birth rate, crude (per 1000 people) - Low & middle income" year 2020.
  • Rate of exclusive breastfeeding in LMICs: 37%. "In low-income and middle-income countries, only 37% of children younger than 6 months of age are exclusively breastfed.” Victora et al. 2016, Abstract.

  Calculation: 6.6 billion total population in LMICs x 1.9% birth rate x 63% not exclusively breastfeeding x $7.76 (average of $14 and $1.51 per child reached [($14 + $1.51)/2= $7.76])= ~$613 million