Lamberti et al.

BMC Public Health 2011, 11(Suppl 3):S15

http://www.biomedcentral.com/1471-2458/11/S3/S15

REVIEW Open Access

Breastfeeding and the risk for diarrhea morbidity

and mortality
Laura M Lamberti1, Christa L Fischer Walker1*, Adi Noiman1, Cesar Victora2, Robert E Black1

Abstract
Background: Lack of exclusive breastfeeding among infants 0-5 months of age and no breastfeeding among
children 6-23 months of age are associated with increased diarrhea morbidity and mortality in developing
countries. We estimate the protective effects conferred by varying levels of breastfeeding exposure against diarrhea
incidence, diarrhea prevalence, diarrhea mortality, all-cause mortality, and hospitalization for diarrhea illness.
Methods: We systematically reviewed all literature published from 1980 to 2009 assessing levels of suboptimal
breastfeeding as a risk factor for selected diarrhea morbidity and mortality outcomes. We conducted random
effects meta-analyses to generate pooled relative risks by outcome and age category.
Results: We found a large body of evidence for the protective effects of breastfeeding against diarrhea incidence,
prevalence, hospitalizations, diarrhea mortality, and all-cause mortality. The results of random effects meta-analyses
of eighteen included studies indicated varying degrees of protection across levels of breastfeeding exposure with
the greatest protection conferred by exclusive breastfeeding among infants 0-5 months of age and by any
breastfeeding among infants and young children 6-23 months of age. Specifically, not breastfeeding resulted in an
excess risk of diarrhea mortality in comparison to exclusive breastfeeding among infants 0-5 months of age (RR:
10.52) and to any breastfeeding among children aged 6-23 months (RR: 2.18).
Conclusions: Our findings support the current WHO recommendation for exclusive breastfeeding during the first 6
months of life as a key child survival intervention. Our findings also highlight the importance of breastfeeding to
protect against diarrhea-specific morbidity and mortality throughout the first 2 years of life.

Background In Africa, Asia, Latin America, and the Caribbean, only

The benefits of breastfeeding on infant and child mor- 47-57% of infants less than two months and 25-31% of
bidity and mortality are well documented, with observa- infants 2-5 months are exclusively breastfed, and the
tional studies dating back to the 1960s and 1970s [1-4]. proportion of infants 6-11 months of age receiving any
Studies show that human milk glycans, which include breastmilk is even lower [7].
oligosaccharides in their free and conjugated forms, are Given that diarrheal disease accounts for approxi-
part of a natural immunological mechanism that mately 1.34 million deaths among children ages 0-59
accounts for the way in which human milk protects months and continues to act as the second leading
breastfed infants against diarrheal disease [5]. In addi- cause of death in this age group [8], it is important to
tion, breastfeeding reduces exposure to contaminated quantify the preventive effect of breastfeeding practices
fluids and foods, and contributes to ensuring adequate on diarrhea-specific morbidity and mortality. Very few
nutrition and thus non-specific immunity. Despite evi- individual studies have been designed or powered to
dence supporting the positive and cost-effective health detect the effects of breastfeeding practices on diarrhea-
impacts of exclusive breastfeeding on child survival [6] specific morbidity and mortality for children 0-23
the practice in resource-poor areas of the world is low. months of age in resource-limited settings.
In 2001, a systematic review of sixteen independent
* Correspondence: cfischer@jhsph.edu studies conducted by the WHO attempted to resolve
the “weanling’s dilemma” in developing countries.
1
Department of International Health, Johns Hopkins Bloomberg School of
Public Health, Baltimore, MD, USA
Full list of author information is available at the end of the article
The review, which assessed the effects of exclusive

© 2011 Lamberti et al; licensee BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons
Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.
Lamberti et al. BMC Public Health 2011, 11(Suppl 3):S15 Page 2 of 12
http://www.biomedcentral.com/1471-2458/11/S3/S15

breastfeeding for 6 months versus 3-4 months with months of age. We excluded studies reporting diarrhea
mixed breastfeeding thereafter, resulted in the recom- as a result of only one microbial cause, and those with
mendation to promote exclusive breastfeeding for the unclear methodology or data in a form that could not be
first 6 months of life [9]. More recently, the authors of extracted for meta-analysis. We also excluded studies
the Lancet nutrition series published a random effects reporting exclusive breastfeeding for children beyond 6
meta-analysis estimating the increased risk of diarrhea- months of age and those failing to restrict the allocation
specific morbidity and mortality among children of diarrhea outcomes to concurrent breastfeeding status.
younger than 2 years in relation to suboptimal breast- Additionally, we excluded morbidity studies with diar-
feeding practices [7]. While these estimates provide con- rhea recall beyond two weeks and mortality studies
firmation of the protective effect of breastfeeding, they where the removal of deaths occurring within the first
were based on a limited data set, rather than a complete three to seven days of life was not possible. For studies
systematic review, and thus a more thorough and reporting outcomes stratified by HIV status, we only
updated revision is warranted. abstracted data on HIV-negative infants and children.
Building upon previous reviews, this systematic review We abstracted data for each diarrhea outcome by
and meta-analyses use carefully developed and standar- breastfeeding exposure levels, which were classified
dized methods to focus on the effects of breastfeeding according to current WHO definitions (Table 1) [12,13].
practices as they relate to diarrhea incidence, prevalence, To allow for the comparability of breastfeeding labels
mortality and hospitalization among children 0-23 and definitions derived from studies published over mul-
months of age. Here we present a comprehensive sys- tiple decades, during which time breastfeeding defini-
tematic review and meta-analysis as evidence to be uti- tions and terms evolved, we assigned the exposure
lized by the Lives Saved Tool (LiST) to model the effect categories described by each study to a WHO category
of breastfeeding practices on diarrhea-specific morbidity on the basis of the study’s definition of that exposure
and mortality [10,11]. The results of our analysis will category, not the authors’ category label. The majority
serve as the basis for generating projections of child of discrepancies between breastfeeding label and defini-
lives that could be saved by increasing exclusive breast- tion arose over the term ‘exclusive breastfeeding’. By
feeding until 6 months of age and continued breastfeed- current standards, ‘exclusive breastfeeding’ does not
ing until 23 months of age. include the ingestion of anything other than breastmilk
and prescribed vitamins and medications, and infants
Methods receiving non-nutritive liquids, such as waters and teas,
We systematically reviewed all literature published from are classified as ‘predominantly breastfed’ [12]. This dis-
1980 to 2009 to identify studies with data assessing tinction was not formally recommended until 1988
levels of suboptimal breastfeeding as a risk factor for when a meeting of the Interagency Group for Action on
diarrhea morbidity and mortality outcomes. We con- Breastfeeding first proposed the development of a set of
ducted our initial search on July 28, 2009 and two standardized breastfeeding definitions [14]. WHO offi-
updated searches on April 8 and May 5, 2010. All cially integrated indicators differentiating between exclu-
searches were completed in Pubmed, EMBASE, the Glo- sive and predominant breastfeeding in 1991 [12]. As
bal Health Library Global Index and Regional Index, such, for this review we assumed the ‘exclusive breast-
and the Cochrane central register for controlled trials feeding’ category was more appropriately labelled ‘pre-
using combinations of key search terms: breastfeeding, dominant breastfeeding’ for studies published prior to
breast milk, human milk, diarrhea, gastroenteritis, mor- 1991, unless the study specifically defined exclusive
bidity, mortality, infant and child. To ensure the identi- breastfeeding according to the current definition.
fication of all relevant literature, we also reviewed the For studies that grouped exclusively and predomi-
references of included papers. nantly breastfed infants into a ‘fully breastfeeding’ cate-
After initially screening for eligibility based on title and gory, we employed a conservative approach in which
abstract, we thoroughly reviewed full publications for fully breastfeeding exposure was treated as predominant.
inclusion and exclusion criteria outlined a priori. We We excluded studies that combined exposures other
included randomized controlled trials (RCT), cohort and than exclusive and predominant breastfeeding into one
observational studies that assessed suboptimal breast- breastfeeding category.
feeding as a risk factor for at least one of the following In this review we did not seek to address the issue of
outcomes: diarrhea incidence, diarrhea prevalence, diar- early initiation of breastfeeding and prelacteal feeds.
rhea mortality, all-cause mortality, and diarrhea hospitali- Thus, in assigning breastfeeding exposure, we did not
zations. Included studies were published in any language differentiate between exclusive and predominant breast-
from 1980 - 2009 and were conducted in developing feeding on the basis of receipt of prelacteal feeds during
countries with a target population of children 0-23 the first 3 days of life.
Lamberti et al. BMC Public Health 2011, 11(Suppl 3):S15 Page 3 of 12
http://www.biomedcentral.com/1471-2458/11/S3/S15

Table 1 Breastfeeding exposures

Exposure Category [12] Permitted to Receive
Exclusive Breastfeeding • breast milk from mother or wet nurse or expressed breast milk
• NO other liquids or solids except vitamin drops or syrups, mineral supplements, or prescribed medicines
Predominant Breastfeeding • breast milk from mother or wet nurse or expressed breast milk
• water and water-based drinks
• NO food-based fluid with the exception of fruit juice and sugar water
• vitamin drops or syrups, mineral supplements, or prescribed medicines
Partial Breastfeeding • breast milk from mother or wet nurse or expressed breast milk
• any other liquids or non-liquids, including both milk and non-milk products
No Breastfeeding • formula and/or animal’s milk
• NO breast milk
Any Breastfeeding • breast milk from mother or wet nurse or expressed breast milk
• Includes children exclusively, predominantly, fully, and partially breastfed

We extracted effect measures and 95% confidence causation bias results when breastfeeding cessation is a
intervals from all included studies. In cases where rela- direct consequence of diarrheal illness. Self-selection
tive risk (RR) was not reported, we generated RR and bias occurs when children are weaned because they
95% confidence intervals using reported numerators and became repeatedly ill or grew improperly while
denominators. breastfed. Although, it has been reported that self-selec-
We organized data into the following age strata: 0-28 tion or reverse causation can also create bias in the
days, 0-5 mos, 0-11 mos, 6-11 mos, 12-23 mos, and 6- opposite direction, with some mothers less likely to
23 mos. We excluded studies with overarching age cate- wean sick children [17]. These biases can be reduced by
gories that could not be collapsed; however, we included the following four methods: (1) exclusion of deaths or
one diarrhea mortality study grouping children 12-35 episodes occurring within the first 7 days of life; (2)
mos and applied its RR to the 12-23 mos analysis [15]. exclusion of infants and young children from non-sin-
For infants aged 0-5 mos, we generated pooled effect gleton and/or premature births and those with low birth
measures using exclusive, predominant, and partial weight, congenital abnormalities, and any other serious
breastfeeding as reference categories. For infants in the illnesses unrelated to the outcome of interest; (3) identi-
0-11 mos category, we used partial and any breastfeed- fication of breastfeeding exposure immediately prior to
ing as reference categories, and for all age categories the onset of illness or mortality as opposed to that con-
extending from 6 or 12 months, we used any breastfeed- current with outcome; (4) assessment of whether wean-
ing as the only reference category. ing was a direct consequence of illness or poor growth
We conducted fixed effects meta-analyses to com- and exclusion of such infants or young children if their
bine effect measures within a given study that had inclusion significantly changes the effect measure [18].
been reported separately for ages falling within the Under our scoring system, we assigned a study 0.5-1
same category in our analysis. To generate a combined point for failure to incorporate each of these four meth-
effect measure across studies, we ran a random effects ods, such that reverse causality was considered not
meta-analysis for each comparison. All meta-analyses likely, likely, and highly likely for studies with zero, 0.5-
were performed using the meta command in STATA 2 and 2.5-4 points, respectively. The studies and the
10.1 [16]. data extracted from each as well as details on scoring
For each outcome of interest, we summarized the evi- studies for reverse causality are available in additional
dence by conducting an assessment of study quality and file 1.
quantitative measures as per CHERG guidelines. As per
the CHERG grading system, the overall quality of evi- Results
dence for each effect estimate receives a score on a four The systematic literature review yielded 2375 unique pub-
point continuum (‘high’, ‘moderate’, ‘low’, ‘very low’), lications, 71 of which contained data on suboptimal
which is then used to either support or oppose its inclu- breastfeeding as a risk factor for the identified outcomes
sion in the LiST model [11]. To further evaluate the of interest (Figure 1). A total of 18 studies met all inclu-
limitations of included studies, we created a scoring sys- sion, exclusion, and analytical criteria and were included
tem to assess the degree to which studies had accounted in the analysis [15,19-35]. Of these, 11 were prospective
for reverse causality and self-selection—two major forms cohort, 4 were cross-sectional observational, and 3 were
of bias in assessing the association between breastfeed- case-control studies. The majority were conducted in
ing and diarrhea morbidity and mortality. Reverse Latin America (n=7) but also took place in Africa (n=4),
Lamberti et al. BMC Public Health 2011, 11(Suppl 3):S15 Page 4 of 12
http://www.biomedcentral.com/1471-2458/11/S3/S15

Figure 1 Synthesis of study identification in review process of the effects of suboptimal breastfeeding exposure on diarrhea
incidence, prevalence, mortality, hospitalizations, and all-cause mortality.

South Asia (n=5), the Middle East (n=2) and the Western studies comparing diarrhea prevalence among exclu-
Pacific (n=2) regions, with one study reporting three dif- sively and suboptimally breastfed neonates (Table 4).
ferent study locations. The numbers of studies included in
each meta-analysis are listed in Tables 2, 3, 4. Diarrhea mortality
In comparison to exclusive breastfeeding, predominant
Diarrhea incidence (RR: 2.28), partial (RR: 4.62) and not (RR: 10.52) breast-
Among infants 0-5 mos of age (Table 2), predominant feeding led to an elevated risk of diarrhea mortality
(RR: 1.26), partial (RR: 1.68) and not breastfeeding (RR: among infants 0-5 mos of age (Table 2; Figures 5, 6, 7).
2.65) resulted in an excess risk of incident diarrhea in Among infants 0-11 mos of age (Table 2), the estimated
comparison to exclusive breastfeeding (Figures 2, 3). risk of diarrhea mortality was higher in partially (RR:
Similarly, the estimated relative risk of incident diarrhea 4.19) and not (RR: 11.73) breastfed infants as compared
was elevated when comparing not breastfed (RR: 1.32) to those predominantly breastfed. For infants and young
to breastfed infants 6-11 mos of age (Table 3; Figure 4). children 6-23 mos of age (Table 3), not breastfeeding
No studies reported diarrhea incidence comparing (RR: 2.18) resulted in an excess risk of diarrhea mortal-
exclusive breastfeeding to suboptimal feeding among ity as compared to breastfeeding (Figure 8). There were
neonates. no studies comparing the outcome of diarrhea mortality
in exclusively versus suboptimally breastfed neonates
Diarrhea prevalence (Table 4).
In comparison to exclusively breastfed infants 0-5 mos
of age, the estimated relative risk of prevalent diarrhea All-cause mortality
was statistically significantly elevated in predominantly As compared to exclusively breastfed infants 0-5 mos of
(RR: 2.15), partially (RR: 4.62), and not (RR: 4.90) age (Table 2), the estimated relative risk of all-cause
breastfed infants (Table 2). Among infants and young mortality was statistically significantly elevated among
children 6-23 mos of age (Table 3), not breastfeeding those predominantly (RR: 1.48), partially (RR: 2.84) and
(RR: 2.07) resulted in an excess risk of prevalent diar- not (RR: 14.40) breastfed. The estimated relative risk of
rhea as compared to breastfeeding. There were no all-cause mortality was higher when comparing not
Lamberti et al. BMC Public Health 2011, 11(Suppl 3):S15 Page 5 of 12
http://www.biomedcentral.com/1471-2458/11/S3/S15

Table 2 The effect of suboptimal breastfeeding on selected outcomes during infancy

0-5 months* 0-11 months*
Outcome Reference Predominant Partial Not Partial Not
Category
Diarrhea Incidence Exclusive 1.26 (0.81-1.95) [22] 1.68 (1.03-2.76) 2.65 (1.72-4.07)
[22,23,28] [22,23,28]
Predominant 1.77 (0.82-3.83) 2.08 (1.58-2.72)
[22,26,27] [22,27]
Partial 1.71 (1.38-2.11)
[22,23,27]

Diarrhea Prevalence Exclusive 2.15 (1.81-2.55) 4.62 (2.37-9.00) 4.90 (2.93-8.21)

[22,30,32,34] [22,30,32] [22,32,34]
Predominant 1.46 (0.95-2.26) 2.40 (1.31-4.43)
[22,27,30] [22,27,34]
Partial 2.05 (1.46-2.88)
[22,27]
Any 1.21 (0.95-1.53) [34]

Diarrhea Mortality Exclusive 2.28 (0.85-6.13) [19,20] 4.62 (1.81-11.76) 10.52 (2.79-39.6)
[19,20] [19,20]
Predominant 2.41 (1.21-4.83)[20] 7.88 (2.64-23.46) 4.19 (2.24-7.84) 11.73 (4.71-29.21)
[20] [25,33] [25,33]
Partial 3.26 (1.15-9.25) [20] 1.69 (1.11-2.58)[25]

All-Cause Mortality Exclusive 1.48 (1.14-1.92) 2.84 (1.63-4.97) 14.40 (6.13-33.86)

[19,20,24] [19,20,24] [19,20]
Predominant 1.69 (1.10-2.61)[20] 8.08 (4.45-14.69)
[20]
Partial 4.77 (2.65-8.61) [20]

Diarrhea Exclusive 2.28 (0.08-6.55) [20] 4.43 (1.75-13.84)[20] 19.48 (6.04-62.87)

Hospitalization [20]
Predominant 3.16 (1.42-7.05) 16.41 (4.59-58.69)
[20,29] [20,29]
Partial 3.95 (1.91-8.19) [20]
*Effect reported as RR (95% CI)[Ref].

Table 3 The effect of not breastfeeding on selected outcomes in children 6-23 months of age
6-11 months* 6-23 months* 12-23 months*
Outcome
Diarrhea Incidence 1.32 (1.06-1.63) [22,27] - -

Diarrhea Prevalence 2.63 (1.04-6.65) [22,27,31] 2.07 (1.49-2.88) [21,22,27,31] 1.39 (1.07-1.80) [21,31]

Diarrhea Mortality 1.47 (0.67-3.25) [19,35] 2.18 (1.14-4.16) [15,19,35] 2.57 (1.10-6.01) [15,35]

All-Cause Mortality 5.66 (1.86-17.20) [19] 3.69 (1.49-9.17) [19,21] 2.23 (0.65-7.59) [21]

Diarrhea Hospitalization 6.05 (2.44-14.97) [29] - -

*Effect reported as RR (95% CI)[Ref]; Any breastfeeding is reference category.
Lamberti et al. BMC Public Health 2011, 11(Suppl 3):S15 Page 6 of 12
http://www.biomedcentral.com/1471-2458/11/S3/S15

Table 4 The effect of suboptimal breastfeeding on selected outcomes in neonates

Outcome Reference Category Predominant Partial Not

Diarrhea Incidence Exclusive -
Predominant 1.67 (0.50-5.52) [27] 0.69 (0.09-5.49) [27]
Partial 0.41 (0.05-3.68) [27]

Diarrhea Prevalence Exclusive

Predominant 4.44 (2.42-8.16) [27] 1.83 (0.73-4.60 [27]
Partial 0.41 (0.17-1.00) [27]

Diarrhea Mortality Exclusive

Predominant 1.40 (0.13-15.42) [19]
Partial

All-Cause Mortality Exclusive 1.41 (1.00-1.99) [19,24] 2.96 (0.75-11.69) [19,24] 1.75 (0.30-10.26) [19]
Predominant 1.33 (0.61-2.91) [19] 1.94 (0.59-6.43) [19]
Partial 1.46 (0.40-5.29) [19]
*Effect reported as RR (95% CI)[Ref].

breastfed (RR: 3.69) to breastfed infants and young chil- Quality assessment and effect size estimates for LiST
dren 6-23 mos of age (Table 3). Among neonates, pre- In table 5, we report the quality assessment of studies by
dominant (RR: 1.41), partial (RR: 2.96), and no (RR: outcome. Using the CHERG grading system for study
1.75) breastfeeding resulted in elevated risk of mortality design and study quality [11], outcome-specific quality
as compared to exclusive breastfeeding (Table 4). was moderate for all outcomes of interest. Although
reverse causation bias was likely or highly likely in the
Diarrhea hospitalizations majority of studies, outcome-specific findings were lar-
The estimated relative risk of hospitalization for diar- gely consistent with all but two studies confirming the
rhea illness was elevated among predominantly (RR: highly protective effect of exclusive breastfeeding and
2.28), partially (RR: 4.43) and not (RR: 19.48) breastfed any breastfeeding among infants 0-5 mos of age and
infants 0-5 mos of age as compared to those exclusively young children 6-23 mos of age, respectively.
breastfed (Table 2). Among infants 6-11 mos of age Applying the CHERG standard rules, strong evidence
(Table 3), not breastfeeding continued to result in a exists for the reduction of diarrhea incidence and diar-
higher risk of hospitalization for diarrhea when com- rhea mortality by exclusive breastfeeding among infants
pared to any breastfeeding (RR: 6.05). There were no 0-5 mos of age and by any breastfeeding among children
studies reporting diarrhea hospitalizations as an out- 6-23 mos of age. In table 6, we present the final effect
come for neonates (Table 4). size estimates to be entered into LiST.

Figure 2 Forest plot for the effect of partial breastfeeding as compared to exclusive breastfeeding on diarrhea incidence among
infants 0-5 months of age.
Lamberti et al. BMC Public Health 2011, 11(Suppl 3):S15 Page 7 of 12
http://www.biomedcentral.com/1471-2458/11/S3/S15

Figure 3 Forest plot for the effect of not breastfeeding as compared to exclusive breastfeeding on diarrhea incidence among infants
0-5 months of age.

Discussion incidence suggesting that the predominate mechanism

We found a sizable body of evidence for the protective by which breastfeeding reduces diarrhea mortality is
effects of breastfeeding against diarrhea incidence, pre- through the reduction of prolonged episodes.
valence, hospitalizations, diarrhea mortality, and In comparison to the Lancet nutrition series [7], we
all-cause mortality. The results of random effects meta- report effect estimates for two additional outcomes—
analyses of eighteen included studies indicated varying diarrhea prevalence and diarrhea hospitalizations, as
degrees of protection across levels of breastfeeding well as additional estimates for neonates separate from
exposure [15,19-35] . the 0-5 months age category. We also conducted meta-
For all outcomes among infants 0-5 mos of age, the analyses comparing reference groups other than exclu-
protection conferred by exclusive breastfeeding was sive breastfeeding for infants 0-5 months of age. The
incrementally greater than that granted by predominant results of our systematic review closely mirrored the
and partial breastfeeding (Table 2). Our results also con- final data set included in the Lancet nutrition series and
firmed a protective effect of any breastfeeding against all thus report nearly identical effect estimates for the
outcomes among infants 6-23 mos of age. The data for meta-analyses of all-cause mortality for 0-5 mos and 6-
neonates alone are limited in that comparisons to exclu- 23 mos; diarrhea mortality for 0-5 mos; and diarrhea
sive breastfeeding, the WHO recommendation for this incidence for predominantly compared to exclusively
age group, were not reported for four out of the five breastfed infants 0-5 mos of age. We excluded two pre-
identified outcomes of interest. Overall, our estimated viously included studies on the basis of diarrhea recall
effect sizes were large, thus suggesting a protective effect beyond two weeks [36,37], and we included four addi-
of breastfeeding among neonates. tional studies not cited by the Lancet nutrition series
The protection conferred by breastfeeding appears to [15,23,28,35]. This resulted in lower effect estimates
operate via two pathways, decreasing diarrhea incidence than those previously reported for the risk of diarrhea
as well as duration. The effect sizes appear to be larger mortality and incidence in not breastfed children 6-23
for the reduction of diarrhea prevalence as compared to mos of age and for the risk of incident diarrhea in

Figure 4 Forest plot for the effect of not breastfeeding as compared to any breastfeeding on diarrhea incidence among infants 6-11
months of age.
Lamberti et al. BMC Public Health 2011, 11(Suppl 3):S15 Page 8 of 12
http://www.biomedcentral.com/1471-2458/11/S3/S15

Figure 5 Forest plot for the effect of predominant breastfeeding as compared to exclusive breastfeeding on diarrhea mortality
among infants 0-5 months of age.

partially and not breastfed infants 0-5 mos of age. the ORs observed after excluding infants dying within
Although we included one of the three studies included the first week of life (OR: 2.36; 95% CI: 1.44-3.87) or
by the Lancet nutrition series in the estimation of the those at high risk of death due to premature birth, con-
risk of diarrhea incidence among children 6-23 mos genital anomaly, or ill health at the time of interview
[22], we further stratified our results in this age category (OR: 2.44; 95% CI: 1.60-3.74) [24]. Despite observing
and thus report this RR under 6-11 rather than 6-23 substantially higher relative risks before methodologi-
mos. Overall, our results confirm and expand upon the cally accounting for reverse causality, the strong protec-
protective effects of breastfeeding as previously reported tive effect of breastfeeding noted by Victora et al.
by the Lancet nutrition series. persisted following this adjustment [4,18].
Although the majority of studies included in this While the current analysis was limited by a lack of
review did not methodologically account for the possibi- geographic variety by outcome, the geographic diversity
lity of reverse causation, it is highly unlikely that this of the overall analysis was actually quite wide with stu-
potential bias was responsible for the large effect sizes dies taking place in eleven unique countries and in all
and consistent findings observed across all age cate- WHO regions except Europe.
gories and outcomes. This assertion is evidenced by the Additionally, the current analysis was limited in that
comparability of findings before and after adjusting for effect measures from studies publishing raw data or esti-
reverse causality within included studies [4,18,20,24]. mates in a form insufficient for meta-analysis were com-
Repeat analyses excluding all deaths occurring within puted without correcting for potential confounders to
7 days of a feeding assessment did not statistically sig- breastfeeding exposure, such as socioeconomic status.
nificantly alter the effect measures observed by Bahl et Still, we do not expect this to constitute a major limita-
al [20]. Similarly, the adjusted odds ratio (2.40; 95% CI: tion since similar methodology has been used in pre-
1.69-3.40) reported by Edmond et al. was very similar to vious studies and since the direction and magnitude of

Figure 6 Forest plot for the effect of partial breastfeeding as compared to exclusive breastfeeding on diarrhea mortality among
infants 0-5 months of age.
Lamberti et al. BMC Public Health 2011, 11(Suppl 3):S15 Page 9 of 12
http://www.biomedcentral.com/1471-2458/11/S3/S15

Figure 7 Forest plot for the effect of not breastfeeding as compared to exclusive breastfeeding on diarrhea mortality among infants
0-5 months of age.

effect sizes were consistent when comparing studies populations. Though there are numerous studies sug-
with and without controls for confounding. Further- gesting that exclusive breastfeeding during the first 6
more, lack of adjustment for confounding may have mos and continued breastfeeding for the second 6 mos
actually led to an underestimation of the protective decrease mortality among infants born to HIV positive
effect of breastfeeding, since poverty is associated with mothers [40,41], further research is warranted as to
longer breastfeeding duration in many of the developing whether the effect sizes reported here are relevant
country populations included in this analysis [38]. among HIV positive mothers and infants.
The quality assessment resulted in a score of moderate
outcome-specific quality (Table 5). According to Conclusions
CHERG standards, the overall score of moderate quality In conclusion, our data confirm and highlight the
across all outcomes of this analysis indicates that these importance of breastfeeding for the prevention of diar-
data represent the best available estimate of the protec- rhea morbidity and mortality. This review also provides
tive effect of breastfeeding against diarrhea-specific mor- updated risk estimates across age categories. Among
bidity and mortality and can therefore be included in infants 0-5 mos of age, these findings support the
the LiST model with confidence [11]. recommendation for exclusive breastfeeding during the
WHO and UNICEF currently recommend exclusive first 6 months of life as a key child survival intervention.
breastfeeding for the first 6 mos of life with continued Furthermore, results among infants and children beyond
feeding through the first year among HIV positive the first 6 mos of age reveal the importance of contin-
mothers, provided that they or their infants receive ARV ued breastfeeding as a critical intervention to protect
drugs during the breastfeeding period.[39] In this review against diarrhea-specific morbidity and mortality
we did not attempt to quantify the relative risks of alter- throughout the first two years of life. Though we have
native infant feeding practices in HIV positive confidence in the strength of the evidence presented

Figure 8 Forest plot for the effect of not breastfeeding as compared to any breastfeeding on diarrhea mortality among infants 6-23
months of age.
Lamberti et al. BMC Public Health 2011, 11(Suppl 3):S15 Page 10 of 12
http://www.biomedcentral.com/1471-2458/11/S3/S15

Table 5 Quality assessment of studies measuring the association between suboptimal breastfeeding and selected
outcomes
Directness
No of studies Design Limitations Consistency Generalizability to Generalizability
(ref)
population of interest to intervention
of interest
Diarrhea Incidence: moderate outcome-specific quality
5 [22,23,26-28] Cohort/ Reverse causality Consistent and all studies showing benefit of EBF Mostly Latin America EBF not reported
Cross- highly likely or likely among infants 0-5 mos of age and benefit of (-0.5) for neonates
sectional for all 5 studies any BF among children 6-23 mos of age (+1) alone
(-0.5)
Diarrhea Prevalence (1-2 week): moderate outcome-specific quality
7 Cohort/ Reverse causality All but one study showing benefit of EBF among Mostly Asia (-0.5) EBF not reported
[21,22,27,30-32,34] Cross- highly likely or likely infants 0-5 mos of age; all studies showing for neonates
sectional for all 7 studies benefit of any BF among children 6-23 mos of alone
(-0.5) age (+1)
Diarrhea Mortality: moderate outcome-specific quality
6 Cohort/ Reverse causality Consistent and all studies showing benefit of EBF Mostly Asia & Latin EBF not reported
[15,19,20,25,33,35] Case- highly likely or likely among infants 0-5 mos of age and benefit of America (-0.5) for neonates
control for 5 of 6 studies any BF among children 6-23 mos of age (+1) alone
(-0.5)
All-Cause Mortality: moderate outcome-specific quality
4 [19-21,24] Cohort Reverse causality All but one study showing benefit of EBF among Mostly Asia (-0.5)
highly likely or likely infants 0-5 mos of age; all studies showing
for all 4 studies benefit of any BF among children 6-23 mos of
(-0.5) age (+1)
Diarrhea Hospitalizations: moderate outcome-specific quality
2 [20,29] Cohort/ Reverse causality Consistent and all studies showing benefit of EBF Equal amount of data EBF not reported
Case- highly likely or likely among infants 0-5 mos of age and benefit of from Asia, Latin America, for neonates
control for both studies any BF among children 6-23 mos of age (+1) Africa & Eastern alone
(-0.5) Mediterranean

Table 6 Application of standardized rules for choice of final outcome to estimate effect of breastfeeding on the
reduction of diarrhea mortality
Outcome Measures Application of
Standard Rules
0-5 months*
Diarrhea n=3; 1594 events Rule 2: APPLY
Incidence The risk of incident diarrhea is 1.26 (0.81-1.95) for predominant BF; 1.68 (1.03-2.76) for partial BF;
2.65 (1.72-4.07) for not BF as compared to EBF
Diarrhea n=2; 80 events
Mortality The risk of diarrhea mortality is 2.28 (0.85-6.13) for predominant BF; 4.62 (1.81-11.76) for partial
BF; 10.52 (2.79-39.6) for not BF as compared to EBF
6-11 months
Diarrhea n=2; 646 events Rule 2: APPLY
Incidence The risk of incident diarrhea is 1.32 (1.06-1.63) for not BF as compared to any BF
Diarrhea n=2; 84 events
Mortality The risk of diarrhea mortality is 1.47 (0.67-3.25) for not BF as compared to any BF
12-23
months
Diarrhea n=0; use estimate for 6-11 mos: n=2; 646 events Rule 2: APPLY
Incidence The risk of incident diarrhea is 1.32 (1.06-1.63) for not BF as compared to any BF
Diarrhea n=2; 84 events
Mortality The risk of diarrhea mortality is 2.57 (1.10-6.01) for not BF as compared to any BF
*Evaluating events for studies where reference category is EBF.
Lamberti et al. BMC Public Health 2011, 11(Suppl 3):S15 Page 11 of 12
http://www.biomedcentral.com/1471-2458/11/S3/S15

here, continued research will be needed to update the 7. Black RE, Allen LH, Bhutta ZA, Caulfield LE, de Onis M, Ezzati M, Mathers C,
Rivera J: Maternal and child undernutrition: global and regional
effect size estimates as diarrhea and all cause mortality exposures and health consequences. Lancet 2008, 371:243-260.
rates continue to decline in many developing countries. 8. Black RE, Cousens S, Johnson HL, Lawn JE, Rudan I, Bassani DG, Jha P,
This review does not evaluate the effect of breastfeed- Campbell H, Walker CF, Cibulskis R, et al: Global, regional, and national
causes of child mortality in 2008: a systematic analysis. Lancet 2010.
ing promotion strategies or the operational challenges of 9. Kramer MS, Kakuma R: The Optimal duration of exclusive breastfeeding:
inspiring mothers to commit to exclusive breastfeeding A systematic review. Cochrane Database Syst Rev 2002, CD003517.
for the first 6 months and to continued breastfeeding 10. Victora CG: LiST: using epidemiology to guide child survival
policymaking and programming. Int J Epidemiol 2010, 39:650-652.
for the following 18 months. Operations research is 11. Walker N, Fischer-Walker C, Bryce J, Bahl R, Cousens S: Standards for
needed to identify methods for maximizing the effective- CHERG reviews of intervention effects on child survival. Int J Epidemiol
ness of breastfeeding promotion programs and policies 2010, 39(Suppl 1):i21-31.
12. Indicators for assessing breastfeeding practices. In World Health
on behaviour change among mothers. Organization. Volume 91. WHO/CDD/SER; 1991:14.
13. Indicators for assessing infant and young child feeding practices. In
Additional material World Health Organization. Volume 07. WHO/CDD/SER; 2007:14.
14. Labbok M, Krasovec K: Toward consistency in breastfeeding definitions.
Stud Fam Plann 1990, 21:226-230.
Additional file 1: Breastfeeding and Diarrhea The additional file 1 is 15. Molbak K, Gottschau A, Aaby P, Hojlyng N, Ingholt L, da Silva AP:
an excel spreadsheet named Final_Web_Appendix. This file contains two Prolonged breast feeding, diarrhoeal disease, and survival of children in
sheets. The ‘Data Abstraction’ sheet includes all data abstracted from Guinea-Bissau. BMJ 1994, 308:1403-1406.
studies, as well as notes on methodology and limitations. The ‘Reverse 16. Stata Corp. Statistical Software: Release 10.1. College Station, TX: Stata
Causality’ sheet includes the assessment sheet used to systematically Corporation; 2008.
score studies on reverse causation bias. 17. Marquis GS, Diaz J, Bartolini R, Creed de Kanashiro H, Rasmussen KM:
Recognizing the reversible nature of child-feeding decisions:
breastfeeding, weaning, and relactation patterns in a shanty town
community of Lima, Peru. Soc Sci Med 1998, 47:645-656.
Acknowledgements 18. Victora CG: Case-control studies of the influence of breastfeeding on
The study was supported by a grant from the Bill and Melinda Gates child morbidity and mortality: methodological issues. In Breastfeeding
Foundation to the US Fund for UNICEF for the ongoing work of the Child nutrition, infection and infant growth in developed and emerging countries.
Health Epidemiology Reference Group. Newfoundland: ARTS Biomedical Publishers and Distributors;Atkinson SA,
This article has been published as part of BMC Public Health Volume 11 Hanson LA, Chandra RK, .St Johns 1990:405-418.
Supplement 3, 2011: Technical inputs, enhancements and applications of the 19. Arifeen S, Black RE, Antelman G, Baqui A, Caulfield L, Becker S: Exclusive
Lives Saved Tool (LiST). The full contents of the supplement are available breastfeeding reduces acute respiratory infection and diarrhea deaths
online at http://www.biomedcentral.com/1471-2458/11?issue=S3. among infants in Dhaka slums. Pediatrics 2001, 108:E67.
20. Bahl R, Frost C, Kirkwood BR, Edmond K, Martines J, Bhandari N, Arthur P:
Author details Infant feeding patterns and risks of death and hospitalization in the first
1
Department of International Health, Johns Hopkins Bloomberg School of half of infancy: multicentre cohort study. Bull World Health Organ 2005,
Public Health, Baltimore, MD, USA. 2Department of Social Medicine, Federal 83:418-426.
University of Pelotas, Porto Alegre, RS, Brazil. 21. Briend A, Wojtyniak B, Rowland MG: Breast feeding, nutritional state, and
child survival in rural Bangladesh. Br Med J(Clin Res Ed) 1988, 296:879-882.
Authors’ contributions 22. Brown KH, Black RE, Lopez de Romana G, Creed de Kanashiro H: Infant-
LML conducted the systematic review, analysis and led the initial manuscript feeding practices and their relationship with diarrheal and other
preparation. CLFW assisted with the analysis and the manuscript preparation. diseases in Huascar (Lima), Peru. Pediatrics 1989, 83:31-40.
AN contributed to the systematic review, data abstraction, and manuscript 23. Clemens J, Elyazeed RA, Rao M, Savarino S, Morsy BZ, Kim Y, Wierzba T,
preparation. CV and REB provided technical leadership and assisted with the Naficy A, Lee YJ: Early initiation of breastfeeding and the risk of infant
interpretation of the analysis and the final manuscript preparation. diarrhea in rural Egypt. Pediatrics 1999, 104:e3.
24. Edmond KM, Zandoh C, Quigley MA, Amenga-Etego S, Owusu-Agyei S,
Competing interests Kirkwood BR: Delayed breastfeeding initiation increases risk of neonatal
The authors have no competing interests. mortality. Pediatrics 2006, 117:e380-386.
25. Garrido F, Borges G, Cardenas V, Bobadilla JL, Ibarra J, Ruiz-Matus C:
Published: 13 April 2011 Postneonatal mortality caused by diarrhea: a case-control study. Salud
Publica Mex 1990, 32:261-268.
26. Ketsela T, Asfaw M, Kebede D: Patterns of breast feeding in western
References Ethiopia and their relationship to acute diarrhoea in infants. J Trop
1. Gordon JE, Chitkara ID, Wyon JB: Weanling diarrhea. Am J Med Sci 1963, Pediatr 1990, 36:180-183.
245:345-377. 27. Lopez-Alarcon M, Villalpando S, Fajardo A: Breast-feeding lowers the
2. Waterlow JC, Thomson AM: Observations on the adequacy of breast- frequency and duration of acute respiratory infection and diarrhea in
feeding. Lancet 1979, 2:238-242. infants under six months of age. J Nutr 1997, 127:436-443.
3. Bauchner H, Leventhal JM, Shapiro ED: Studies of breast-feeding and 28. Macias-Carrillo C, Franco-Marina F, Long-Dunlap K, Hernandez-Gaytan SI,
infections. How good is the evidence? JAMA 1986, 256:887-892. Martinez-Lopez Y, Lopez-Cervantes M: Breast feeding and the incidence of
4. Victora CG, Smith PG, Vaughan JP, Nobre LC, Lombardi C, Teixeira AM, acute diarrhea during the first three months of life. Salud Publica Mex
Fuchs SC, Moreira LB, Gigante LP, Barros FC: Infant feeding and deaths 2005, 47:49-57.
due to diarrhea. A case-control study. Am J Epidemiol 1989, 29. Mahmood DA, Feachem RG, Huttly SR: Infant feeding and risk of severe
129:1032-1041. diarrhoea in Basrah city, Iraq: a case-control study. Bull World Health
5. Morrow AL, Ruiz-Palacios GM, Jiang X, Newburg DS: Human-milk glycans Organ 1989, 67:701-706.
that inhibit pathogen binding protect breast-feeding infants against 30. Mihrshahi S, Oddy WH, Peat JK, Kabir I: Association between infant
infectious diarrhea. J Nutr 2005, 135:1304-1307. feeding patterns and diarrhoeal and respiratory illness: a cohort study in
6. Riordan JM: The cost of not breastfeeding: a commentary. J Hum Lact Chittagong, Bangladesh. Int Breastfeed J 2008, 3:28.
1997, 13:93-97.
Lamberti et al. BMC Public Health 2011, 11(Suppl 3):S15 Page 12 of 12
http://www.biomedcentral.com/1471-2458/11/S3/S15

31. Mulder-Sibanda M, Sibanda-Mulder FS: Prolonged breastfeeding in

Bangladesh: indicators of inadequate feeding practices or mothers’
response to children’s poor health? Public Health 1999, 113:65-68.
32. Popkin BM, Adair L, Akin JS, Black R, Briscoe J, Flieger W: Breast-feeding
and diarrheal morbidity. Pediatrics 1990, 86:874-882.
33. Victora CG, Smith PG, Vaughan JP, Nobre LC, Lombardi C, Teixeira AM,
Fuchs SM, Moreira LB, Gigante LP, Barros FC: Evidence for protection by
breast-feeding against infant deaths from infectious diseases in Brazil.
Lancet 1987, 2:319-322.
34. Vieira GO, Silva LR, de OVT: Child feeding and diarrhea morbidity. J Pediatr
(Rio J) 2003, 79:449-454.
35. Yoon PW, Black RE, Moulton LH, Becker S: Effect of not breastfeeding on
the risk of diarrheal and respiratory mortality in children under 2 years
of age in Metro Cebu, The Philippines. Am J Epidemiol 1996,
143:1142-1148.
36. Kumar V, Kumar L, Diwedi P: Morbidity related to feeding pattern in
privileged urban and under privileged rural infants. Indian Pediatr 1981,
18:743-749.
37. al-Ali FM, Hossain MM, Pugh RN: The associations between feeding
modes and diarrhoea among urban children in a newly developed
country. Public Health 1997, 111:239-243.
38. Contemporary patterns of breast-feeding: report on the WHO
Collaborative Study on Breast-feeding. In Book Contemporary patterns of
breast-feeding: report on the WHO Collaborative Study on Breast-feeding.
Volume 211. City:Geneva Switzerland: World Health Organization; 1981:24.
39. Guidelines on HIV and infant feeding: principles and recommendations
for infant feeding in the context of HIV and a summary of evidence.
Book Guidelines on HIV and infant feeding: principles and recommendations
for infant feeding in the context of HIV and a summary of evidence City:
Geneva Switzerland: World Health Organization; 2010.
40. Homsy J, Moore D, Barasa A, Were W, Likicho C, Waiswa B, Downing R,
Malamba S, Tappero J, Mermin J: Breastfeeding, mother-to-child HIV
transmission, and mortality among infants born to HIV-Infected women
on highly active antiretroviral therapy in rural Uganda. J Acquir Immune
Defic Syndr 2010, 53:28-35.
41. Coovadia HM, Rollins NC, Bland RM, Little K, Coutsoudis A, Bennish ML,
Newell ML: Mother-to-child transmission of HIV-1 infection during
exclusive breastfeeding in the first 6 months of life: an intervention
cohort study. Lancet 2007, 369:1107-1116.

doi:10.1186/1471-2458-11-S3-S15
Cite this article as: Lamberti et al.: Breastfeeding and the risk for
diarrhea morbidity and mortality. BMC Public Health 2011 11(Suppl 3):S15.

Submit your next manuscript to BioMed Central

and take full advantage of:

• Convenient online submission

• Thorough peer review
• No space constraints or color figure charges
• Immediate publication on acceptance
• Inclusion in PubMed, CAS, Scopus and Google Scholar
• Research which is freely available for redistribution

Submit your manuscript at

www.biomedcentral.com/submit