Roberts et al.

International Breastfeeding Journal (2018) 13:39

https://doi.org/10.1186/s13006-018-0179-4

SHORT REPORT Open Access

Evidence of inflated exclusive breastfeeding

estimates from a clinical trial in Bangladesh
Thomas J. Roberts1* , Yana E. Hoy-Schulz2, Kaniz Jannat3 and Julie Parsonnet2,4

Abstract
Suboptimal breastfeeding is a major cause of infant morbidity and mortality across the world. Inconsistent data has
hampered quantification of this practice, however, limiting breastfeeding promotion efforts. As part of a clinical trial in
Dhaka, Bangladesh, data was collected on breastfeeding patterns among 125 infants. Infants were ages 4 to 12 weeks
(mean = 8.05, SD = 2.13) at the time of enrollment, and breastfeeding data were collected at 24 study visits during a
twelve-week period. Breastfeeding status was assessed using the WHO-recommended “current status” (24-h recall)
method. These data were used to calculate two measures: a longitudinal estimate of exclusive breastfeeding since
birth and a simulated cross-sectional prevalence to approximate common data collection methods. Infants were then
ranked based on their breastfeeding status at all study visits and grouped into quartiles and compared using
hospitalization data recorded for all infants as part of the original study. These data showed large differences in
estimates of exclusive breastfeeding behaviors when assessed longitudinally (8.8% exclusive breastfeeding) vs.
calculating a cross-sectional prevalence (56.2% exclusive breastfeeding). Additionally, when infants were grouped by
quartile of breastfeeding behavior and matched with hospitalization records, it was found that infants in the lowest
quartile of breastfeeding behaviors were significantly more likely to be hospitalized than infants in the highest quartile.
These results provide further evidence that current breastfeeding epidemiology studies may overestimate rates of
exclusive breastfeeding. They also provide further evidence to support the significant infant health benefits from
breastfeeding promotion.
Trial registration: ClinicalTrials.gov NCT01899378. Registered July 10, 2013.
Keywords: Exclusive breastfeeding, Breastfeeding promotion, Measurement error, Epidemiology, Infant nutrition

Background even water—with the exception of oral rehydration solu-

Advocacy for exclusive breastfeeding is one of the most tion or drops/syrups or vitamins, minerals or medicines”
important health interventions for infants and mothers [3]. However, the WHO recommends data collection
in low-income countries [1]. Despite breastfeeding pro- using the “current status” method, a point prevalence
motion efforts, it is estimated that 47 million based on whether infant were given any complementary
disability-adjusted life years were attributable to subopti- foods during the 24 h prior to the time of survey. Many
mal breastfeeding in 2010, making it the 5th greatest national and international surveys collect breastfeeding
cause of human morbidity among 67 measured risk fac- data using this method.
tors [2]. These estimates of the disease burden due to The “current status” data collection method has been
suboptimal breastfeeding are limited by discrepancies previously shown to overestimate the prevalence of ex-
between breastfeeding definitions and measuring meth- clusive breastfeeding when compared to prospectively
odologies. The World Health Organization (WHO) de- collected data [4, 5]. However, it not feasible to collect
fines exclusive breastfeeding as an infant receiving “only prospective data as part of the large cross-sectional sur-
breastmilk. No other liquids or solids are given—not veys frequently used to estimate national and inter-
national breastfeeding trends [6]. Additionally, the
“current status” method may be less inaccurate than
* Correspondence: [email protected]
1
Department of Medicine, Massachusetts General Hospital, 55 Fruit St, methods based on maternal recall since birth [7]. The
Boston, MA 02114, USA aim of this study was to compare prospectively collected
Full list of author information is available at the end of the article

© The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Roberts et al. International Breastfeeding Journal (2018) 13:39 Page 2 of 5

breastfeeding data with a simulated cross-sectional chosen to approximate results reported in recent surveys
sample from the same population of Bangladeshi infants and international breastfeeding epidemiology studies [2, 9].
to provide further information about the discrepancies To assess the association between breastfeeding behav-
between reported breastfeeding rates and actual iors and health outcomes in this sample, infants were
breastfeeding practices. ranked based on their breastfeeding status at all study
visits. Each infant was assigned a value of 0 to 3 for each
study visit based on their breastfeeding status (0 for no
Methods breastfeeding, 1 for partial breastfeeding, 2 for predom-
As part of a randomized controlled trial investigating inate breastfeeding, 3 for exclusive breastfeeding). The
the safety and efficacy of probiotics in infants, we col- mean across all study visits was calculated for each in-
lected infant feeding data on 137 infants less than fant and the infants were ranked highest to lowest based
6 months old [8]. The main study was conducted at the on this value. The infants were grouped based on quar-
International Center for Diarrheal Disease, Bangladesh tile and compared using the hospitalization data.
(icddr,b) in Dhaka, Bangladesh from October 2013
through August 2014. The study enrolled healthy infants
between 4 and 12 weeks of age (mean age of enrollment Results
= 8.05 weeks, SD = 2.13 weeks) from designated neigh- The cross-sectional sample contained 112 infants with a
borhoods near the iccdr,b. They were randomized to one mean age of 13.8 weeks (SD = 4.1 weeks). The sample
of four study groups. Three groups received probiotics size was 112 because breastfeeding data was not re-
at varying frequency (daily, weekly, biweekly) and the corded at some of the randomly selected visits. Using
fourth group served as a control. Data from all hospitali- this sample, we calculated the prevalence of exclusive
zations during each infant’s period of enrollment were breastfeeding among children 2–5 months to be 56.2%.
captured as part of the main study. The main study This number is consistent with data from the 2011
found no difference in health outcomes or breastfeeding Bangladesh Demographic and Health Survey that
behaviors between the study groups [8]. showed an exclusive breastfeeding prevalence of 54.6%
Data were collected at up to 24 study visits per infant for this age group [9]. However, Fig. 1 shows the breast-
over a twelve-week period. The timing and frequency of feeding status for each child at each study visit, and
study visits was based on the study design of the original these data revealed large discrepancies between the
trial. At each study visit, a brief questionnaire was ad- cross-sectional prevalence and actual breastfeeding be-
ministered to the infants’ mother. Breastfeeding data haviors. Of the 125 children in our sample, only 11
were collected using the WHO-recommended “current (8.8%) were exclusively breastfed throughout the study
status” (24-h recall) method. Infants with less than 10 period. In many instances, children went through
study visits with breastfeeding information were ex- multi-day periods of partial or predominant breastfeed-
cluded from this analysis. The final sample size for this ing before returning to “exclusive” status. The
analysis was 125 infants. cross-sectional prevalence using the “current status”
When categorizing each infant’s breastfeeding status at method (56.2%) was 47.4 percentage points higher than
study visits, we used WHO definitions of breastfeeding the actual (8.8%) exclusive breastfeeding prevalence cal-
behaviors [3]. Infants given only breast milk over the culated from prospectively-collected data.
previous 24 h were considered exclusively breastfed. In- When infants were grouped by quartile based on
fants given only breast milk and water over the previous their breastfeeding status and compared using
24 h were considered predominately breastfed. Infants hospitalization records, no infants in the top quartile
given breast milk and any other complementary foods were hospitalized during the study period. Three in-
over the past 24 h were considered partially breastfed. fants (4.8%) in the middle quartiles, and five infants
The longitudinal prevalence of exclusive breastfeeding (16.1%) in the lowest quartile were hospitalized (Fig. 1;
was determined using the WHO definition of “no other Fisher Exact Test, p value < 0.05). Reverse causality (ie
food or drink, not even water, except breast milk for sick infants may be unable to breastfeed) could not be
6 months of life.” To be considered “exclusively breast- completely excluded. However, the available data from
fed” longitudinally, the infant must have been given safety reports and hospitalization records indicated that
nothing but breast milk at every prior 24-h recall. the causes of suboptimal breastfeeding were often inde-
To simulate data from a cross-sectional survey, a sin- pendent of the infants’ baseline health status. For ex-
gle study visit was randomly selected for each infant and ample, in one case breastfeeding was prematurely
the breastfeeding status at that visit was recorded as that stopped when the mother traveled away from the home.
infant’s breastfeeding status. The metric of prevalence of In another case breastfeeding was not initiated because
exclusive breastfeeding among infants 2–5 months was the infant was adopted.
Roberts et al. International Breastfeeding Journal (2018) 13:39 Page 3 of 5

Not breastfed
Partial BF
Predominant BF
Exclusive BF kno wn
BF status not kno wn
BF status not kno wn
Hospitalized: diarrhea
Hospitalized: pneumonia

Fig. 1 Breastfeeding status at each study visit by age of the infants. Each cell represents one study visit. The frequency of study visit was determined
by the study design of the original trial. Data are listed by age of infant, not the actual date of collection. The final column indicates whether infant
was hospitalized, including the reported reason for the hospitalization, during the study period

Conclusions populations [5, 7], it appears that only a minority of

These data provide further evidence of the large discrep- women thought to be exclusively breastfeeding actually
ancy between the prevalence of breastfeeding reported achieves this goal. Women in this study received educa-
using the “current status” method and actual breastfeed- tion on the importance of breastfeeding as part of
ing behaviors observed using prospectively collected participation in the study but provided other foods
data. Based on these results and similar studies in other nevertheless. The disconnect between education and
Roberts et al. International Breastfeeding Journal (2018) 13:39 Page 4 of 5

behavioral change highlights areas for further study to donors to their research efforts and is thankful to the Governments of
improve the effectiveness of breastfeeding interven- Bangladesh, Canada, Sweden and the UK for providing core/unrestricted
support. Funders had no role in the study design, data collection, data analysis,
tions in developing countries. They also reveal how decision to publish, or preparation of the manuscript.
much remains to be done to achieve infant nutrition
goals. Availability of data and materials
The datasets used and analyzed during the current study are available from
This report was not able to assess the effects of the corresponding author on reasonable request.
prelacteal feeds as data about feeding practices during
the infant’s first 24 h were not collected during the Authors’ contributions
TR: Conception, study design, data collection, data analysis, data
main study. However, breastfeeding practices through interpretation, literature search, writing. YEH-S: Study design, data collection,
the first 6 months have been shown to be independ- data analysis, writing. KJ: Study design, data collection, data analysis, writing.
ently associated with health outcomes [10–12], and JP: Conception, study design, data analysis, writing. All authors read and ap-
proved the final manuscript.
we feel that this report makes valuable contributions
to the breastfeeding literature by further characteriz- Ethics approval and consent to participate
ing discrepancies between different data collection Ethics approval was received from Stanford University’s Institutional Review
Board (Protocol #25487) and icddr,b’s Ethical Review Committee (Protocol
methods in a South Asian population. Additionally, PR-13022). Informed consent was received from all study participants.
this study is limited by the relatively small sample
size. However, the consistency of these data with data Consent for publication
Not Applicable.
from studies in other contexts and larger contempor-
aneous studies indicates they reflect true breastfeeding Competing interests
behaviors and accurately represent the discrepancies All authors declare they have no competing interests.

between data collection methods.

The discrepancies presented here are concerning. Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in
However, it is not feasible for large cross-sectional published maps and institutional affiliations.
studies to prospectively collect data throughout the
first 6 months of infants’ lives. These changes would Author details
1
Department of Medicine, Massachusetts General Hospital, 55 Fruit St,
limit the sample sizes of the surveys and the fre- Boston, MA 02114, USA. 2Department of Medicine, Stanford University School
quency of their administration, jeopardizing valuable of Medicine, Stanford, USA. 3Environmental Intervention Unit, Infectious
sources of breastfeeding data. As the body of evi- Disease Division, International Centre for Diarrheal Disease Research, Dhaka,
Bangladesh. 4Department of Health Research and Policy, Stanford University
dence quantifying differences between data collection School of Medicine, Stanford, USA.
methods continues to grow, future research may be
able to propose correction factors that allow re- Received: 24 January 2018 Accepted: 30 July 2018
searchers to more reliably compare data collected
using different methods and improve estimates of ac- References
tual breastfeeding behaviors. 1. Bhutta ZA, Das JK, Rizvi A, Gaffey MF, Walker N, Black RE, et al. Evidence-
based interventions for improvement of maternal and child nutrition: what
can be done and at what cost? Lancet. 2013;382:452–77.
Abbreviations 2. Lim SS, Vos T, Flaxman AD, Danaei G, Shibuya K, Adair-Rohani H, et al. A
BF: breastfeeding; DHS: Demographic and Health Survey; comparative risk assessment of burden of disease and injury attributable to 67
icddr,b: International Center for Diarrheal Disease Research, Bangladesh; risk factors and risk factor clusters in 21 regions, 1990-2010: a systematic
WHO: World Health Organization analysis for the global burden of disease study 2010. Lancet. 2012;380:2224–60.
3. World Health Organization: Exclusive breastfeeding for optimal growth,
Acknowledgements development and health of infants. 2018. http://www.who.int/elena/titles/
The authors with to thank all of the individuals at the International Center exclusive_breastfeeding/en. Accessed 3 July 2018.
for Diarrheal Disease Research, Bangladesh (icddr,b), particularly Leanne 4. Aarts C, Kylberg E, Hornell A, Hofvander Y, Gebre-Medhin M, Greiner T. How
Unicomb, Mostafizur Rahman, and the team of research assistants, involved exclusive is exclusive breastfeeding? A comparison of data since birth with
with the previously published clinical trial. We would like to thank Dr. Steve current status data. Int J Epidemiol. 2000;29:1041–6.
Luby for his guidance with the experimental design, data collection, and 5. Binns C, Hewitt K. Reported breastfeeding rates in the Asia-Pacific region.
analysis of the previously published clinical trial. We would also like to Saira Curr Pediatr Rev. 2012;8:339–45.
Zaidi for her assistance with the previously published clinical trial. 6. Roberts TJ, Carnahan E, Gakidou E. Can breastfeeding promote child health
equity? A comprehensive analysis of breastfeeding patterns across the
Funding developing world and what we can learn from them. BMC Med. 2013;11:254.
This study was supported by the Thrasher Research Fund Early Career Award 7. Bland RM, Rollins NC, Solarsh G, Van den Broeck J, Coovadia HM, Child
(YEHS), Freeman Spogli Institute Underdevelopment Grant (JP), Stanford Center Health Group. Maternal recall of exclusive breast feeding duration. Arch Dis
for Innovation in Global Health Seed Grant (JP), Global Health Equity Scholars Child. 2003;88:778–83.
Fellowship (YEHS), Stanford Medical Scholars Fund (TR), Stanford Child Health 8. Hoy-Schulz YE, Jannat K, Roberts TJ, Zaidi SH, Unicomb L, Parsonnet J, et al.
Research Institute Postdoctoral Grant through Lucile Packard Foundation for Safety and acceptability of lactobacillus reuteri DSM 17938 and
Children’s Health and the Stanford CTSA (UL1 RR025744) (YEHS), the TL1 Clinical Bifidobacterium longum subspecies infantis 35624 in Bangladeshi infants: a
Research Training Program of the Stanford Clinical and Translational Science phase I randomized clinical trial. BMC Complement Altern Med. 2016;16:44.
Award to Spectrum (NIH TL1 TR 001084) (YEHS), and the T32 Epidemiology of 9. Bangladesh Demographic and Health Survey 2011. Dhaka, Bangladesh &
Emerging Infectious Disease and Bioterrorism training grant (5T32AI052073–08) Calverton, Maryland, USA: National Institute of population research and
(YEHS). The ICDDR,B acknowledges with gratitude the commitment of the training (NIPORT), Mitra and Associated, and ICF International; 2013:161–177.
Roberts et al. International Breastfeeding Journal (2018) 13:39 Page 5 of 5

10. Arifeen S, Black RE, Antelman G, Baqui A, Caulfield L, Becker S. Exclusive

breastfeeding reduces acute respiratory infection and diarrhea deaths
among infants in Dhaka slums. Pediatrics. 2001;108:e67.
11. Lamberti LM, Walker CLF, Noiman A, Victora C, Black RE. Breastfeeding and the
risk for diarrhea morbidity and mortality. BMC Public Health. 2011;11:S15.
12. Lamberti LM, Zakarija-Grković I, Walker CLF, Theodoratou E, Nair H, Black RE,
et al. Breastfeeding for reducing the risk of pneumonia morbidity and
mortality in children under two: a systematic literature review and meta-
analysis. BMC Public Health. 2013;13:S18.