Peer counselling improves breastfeeding practices: A cluster randomized controlled trial in urban Bangladesh

Abstract This study aimed to evaluate the impact of peer counselling on early initiation of breastfeeding (EIBF) and exclusive breastfeeding (EBF) rates for mother–infant pairs living in urban slums, Dhaka, Bangladesh. This randomized controlled trial enrolled 350 mother–infant pairs from selected slums between September 2014 and July 2016. The women assigned to intervention group received peer counselling from locally recruited, trained community female volunteers starting in third trimester of pregnancy until 6 months after delivery; control group received no intervention. EIBF, any liquids given after birth, and EBF were compared between groups. Follow‐up was scheduled at enrolment, following childbirth, and every 2 months up to 6 months after delivery. Multiple logistic regressions were used to assess the effect peer counselling and other associated factors on EIBF and EBF practices. EIBF rate was higher in the intervention group than in the control group (89.1% vs. 77.4%, p = .005). More mothers in intervention group were exclusively breastfeeding at 5 months than mothers in the control group (73% vs. 27%, p < .005). Control mothers were twice as likely to not practice EIBF compared with intervention mothers (adjusted odds risk [aOR]: 2.53, CI [1.29, 4.97], p = .007). Overall, caesarean section was associated with an 8.9‐fold higher risk of not achieving EIBF (aOR: 8.90, CI [4.05, 19.55], p < .001). Intervention mothers were 5.10‐fold more likely to practice EBF compared with control mothers (aOR: 5.10, CI [2.89, 9.01], p < .001) at 5 months. This study demonstrates peer counselling can positively influence both EIBF and EBF among mothers living in urban area.

and exclusive breastfeeding (EBF) is a key intervention. Early initiation of breastfeeding (EIBF) is defined as putting the infant to the breast within 1 hr of birth. EBF is recommended by the World Health Organization (WHO), which states that infants should be breastfed exclusively from birth until 6 months of age, followed by continued breastfeeding alongside gradual introduction of solid foods (World Health Organization and UNICEF, 2003). Several beneficial effects of breastfeeding have been reported, including reductions in the incidence and severity of infectious diseases such as diarrhoea, respiratory tract infections, otitis media, and urinary tract infections, as well as reductions in the incidence of Types 1 and 2 diabetes mellitus, being overweight, obesity, and asthma (Gartner et al., 2005). Conversely, early introduction of breast milk substitutes or semi-solid foods and delayed introduction of appropriate semi-solid complementary foods are risk factors associated with the rapid increase in the prevalence of undernutrition among children between 6 and 24 months old (Ramachandran, 2004 Edmond et al., 2006). Additionally, the risk of neonatal mortality was higher among infants without EBF practices compared with those with EBF. Improved breastfeeding during the neonatal period helps to reduce mortality and benefits the health, growth, and development of the child in the first year and beyond (Huffman, Zehner, & Victora, 2001). Establishment of good breastfeeding practices in the first days after birth is critical for successful breastfeeding practice and infant health. Initiation of breastfeeding is easiest and most successful when a mother is physically and psychologically prepared for birth and is informed, supported, and confident in her ability to breastfeed and care for the newborn (Development., A.f.E., 2003).
The importance of community support for EIBF was noted in a recent neonatal survival review published inThe Lancet covering a range of potential interventions (including EIBF; Darmstadt et al., 2005).
Community support via outreach and health education is important for the prevention of neonatal deaths in settings with high mortality and weak health systems to improve home-care practices and increase the demand for skilled care and care seeking (Darmstadt et al., 2005).
Community-based interventions such as one-to-one or group counselling to promote and support infant and young child feeding, along with policy measures and improvements in facility-based services, are strongly emphasized in the global strategy developed jointly by the WHO and United Nations Children's Fund (UNICEF). A recent metaanalysis revealed delivery of individual peer counselling to mothers significantly increased the rates of EBF in both the neonatal period (15 studies; odds ratio [OR] = 3.45, 95% CI [2.20, 5.42], p < .0001) and at 6 months of age (nine studies; 1.93, [1.18, 3.15], p < .0001 ;Edmond et al., 2006). Studies conducted in various developing countries have shown the extent of malnutrition is more severe among children living in slums compared with children living in developed city areas and, sometimes, even compared with children living in rural regions (Fakir & Khan, 2015;Glewwe, Koch, & Nguyen, 2002). An increased risk of inappropriate child feeding practices has been reported among the rapidly growing slum populations in urban areas (Kumar, Nath, & Reddaiah, 1989), as many of these families lack the traditional support of the joint family system. Urban slums are a continually increasing phenomenon in South Asian countries and are underserved by medical facilities (Parvin, Ahsan, & Shaw, 2013). Therefore, we undertook a randomized controlled trial (RCT) to examine whether peer counselling can improve EIBF and EBF practices in an urban slum in Dhaka, Bangladesh.

| Trial design
This prospective, RCT to examine the impact of peer counsellingstarting in the third trimester of pregnancy and continuing until 6 months after delivery-on infant feeding practices was approved by the local ethics committee and registered at clinicaltrials.gov as NCT03040375 (last updated: January 31, 2017). Study participants were randomized into two groups: peer counselling (intervention) and no peer counselling (control).

| Setting
The study area was Mirpur, a district of Dhaka with a total population of 5,580,000. The major source of income in Mirpur is wage labour (49.9%), and 53.0% of household heads have no education. The most common improved source of drinking water is a piped water connection inside the user's dwelling, plot or yard (62.8%), and 37.5% of the

| Study participants
Pregnant women aged between 16 and 49 years old with no more than three living children or a parity of five and who intended to reside in the area for at least 6 months after delivery were identified for the study from a house-to-house survey. Women with documented heart disease, insulin-dependent diabetes mellitus, or pre-eclampsia in a previous pregnancy were not included. Mother-infant pairs were excluded in cases with congenital anomalies, admission to intensive care, or a birthweight below 1.5 kg (Haider, Ashworth, Kabir, & Huttly, 2000).

Key messages
• Peer counselling has a positive effect on initiation of breastfeeding within 1 hr of birth and exclusive breastfeeding practices.
• Emphasis on encouraging EIBF and discouraging prelacteal feeding should be focused on mothers living who delivered by caesarean section.

| Ethics approval and consent to participate
This study (PR-14091) was approved by the Research Review Committee and Ethical Review Committee, the two compulsory components of the institutional review board of International Centre for Diarrhoeal Disease Research, Bangladesh (icddr,b). Written consent was provided by all study participants. The field research assistants informed the participants about the purpose of the study at the beginning of each interview by reading a consent form. Written consent was also taken from the parent/guardian for anthropometric measurement of the study infants. The respondents were informed of the important point that their participation was voluntary and they were allowed to withdraw themselves at any point of time during the interview/study. The respondents were also informed that the information obtained from this survey would be anonymous and have a broader impact, guide the development of policies and programs related to child nutrition, growth, and development, and would contribute to improve the child health services in Bangladesh and elsewhere.

| Selection and training of peer counsellors
Women with personal breastfeeding experience, who had at least 8 years of schooling, who were motivated to help other mothers breastfeed, and who resided in the intervention area were selected to become peer counsellors. Peer counsellors were recruited by the International Centre for Diarrhoeal Disease Research, Bangladesh (icddr,b) and paid a monthly salary. The WHO/UNICEF Breastfeeding Counselling Course, which was validated in a previous study (Haider, Ashworth, Kabir, & Huttly, 2000), was adapted to the local language and culture and used to train peer counsellors. Training was provided over 40 hr (4 hr daily for 10 days). Counselling skills were taught by demonstration and role play, including listening to mothers, learning about their difficulties, assessing the position and attachment of the baby during breastfeeding, building mothers' confidence, giving support, and providing relevant information and practical help when required. One peer counsellor was recruited from each intervention cluster and was responsible for delivering the intervention to 30-35 mothers residing in the same area. Their performance was monitored at least three times over the study period by one breastfeeding supervisor based at the icddr,b field office.

| Counselling
The intervention group received at least 10 scheduled visits between the third trimester and 6 months after delivery: three in the last trimester of pregnancy, three in the first month after delivery (one within 48 hr of delivery, one 10-14 days after, and one 24-28 days after), and four visits in the second to sixth months. Counsellors were free to make and receive additional contact with the intervention group if the mothers required additional support. Counselling was given at home, and other family members were included. The duration of each visit was typically 20-40 min. During the two antenatal contacts, the peer counsellors emphasized to the mothers and other members of the family who would support her during delivery the importance of the mother holding the baby within a few minutes of delivery and gave instructions on how to initiate breastfeeding within 1 hr of delivery. They discouraged prelacteal feeds and the use of other fluids and foods after lactation was initiated. The peer counsellors encouraged the mothers to eat more of their usual foods to support lactation and to rest appropriately during the third trimester. These meetings also covered problems with breastfeeding that the mother may encounter and how best to address them.
Fathers and other household members were briefed on the importance of keeping the mother happy and joyful and not subjecting her to violence or harsh treatment. Fortnightly group meetings were organized with the participation of pregnant women, lactating mothers, and their family members. In these meetings, mothers and family members were encouraged to take extra care of the pregnant woman to ensure proper nutrition, attendance of antenatal check-ups, proper intake of iron tablets, safe delivery, delivery at health facilities, colostrum feeding, and EIBF and continuation of EBF. The peer counsellors emphasized the negative effects of prelacteal feeding, breast milk substitutes, and early introduction of complementary feeding. The peer counsellors provided frequent home visits to support the mothers with EBF.12 2.6 | Sample size The number of mother-infant pairs required was calculated based on achieving a difference of least 20 percentage points in the prevalence of EBF between the intervention and control groups, with a 5% significance level, 80% power, and design effect of 1.5 in a two-tailed test. The expected prevalence of EBF in the control group was 27% (from unpublished data from another RCT in Mirpur area, personal communication with M. J. Dibley et al.). A sample size of 156 mother-infant pairs was calculated; due to the nature of community-level interventions, expected migration rate of 12%, and potential loss to follow-up, the estimated sample size was increased to 175 mother-infant pairs for each group.

| Randomization
The trial was conducted in two wards of Mirpur municipality. The average population of a ward is 350,000. There are approximately five wards, each with an average population of 70,000. The interviewer visited door-to-door to identify pregnant women in their third trimester at every 2 months interval. Women who fulfilled the inclusion criteria and provided informed consent were randomly allocated to either the intervention or control group.

| Collection of data
Four female field research assistants were recruited and trained over 2 weeks. Each mother was visited by an interviewer four times.
Structured questionnaires with precoded closed questions were used during the home interviews. Data on socio-economic and demographic variables, maternal and pregnancy factors, and previous infant feeding were obtained at enrolment. Details of delivery and early feeding were obtained within 72 hr of birth (60-80 hr). Data on feeding practices were collected every 2 months. To ensure the quality of the data, 10% of the interviews were observed by the field research supervisor. The data were separately recorded, entered, and analysed to ensure consistency with the data collected by the field research assistants. If any inconsistencies were observed, the principal investigator revisited and re-interviewed the same subjects using the same questionnaire. Unscheduled field visits and spot checks by the principal investigator and research physician were also performed to monitor the quality of data collection.

| Data analysis
Data were entered into STATA v13 (StataCorp; College Station, TX, USA). The primary analyses compared the prevalence of EIBF and EBF in children at 0-6 months of age using Pearson's chi square test to calculate 95% confidence intervals for the group difference.
Secondary outcome variables included the proportion of mothers who fed colostrum or prelacteals after delivery. For the primary trial outcome, the results of two-sided 5% tests are reported.
Bivariate and multivariable logistic regressions were used to identify predictive variables, and ORs with 95% confidence intervals and value were used to measure the strength of the associations.
Variables with significant associations (aOR, 95% confidence intervals [CI], p value < .05) were included in the final regression model; variables with a p value < .05 were defined as predictor variables.

| Baseline comparison
Of 700 pregnant women contacted, 378 were enrolled in the study. Of the 322 (46%) women excluded, the main reasons were that it was too early in the pregnancy, the mother did not live in the Assessed for eligibility (n=700) Excluded (n=322)

Follow up of mother infant pair at 3rd month (131)
Lost of follow up=44(25%)

Follow up of mother infant pair at 3rd month (140)
Lost of follow up=26(16%)

Total births (170)
Live births (166) Stillbirths (2) (Table 1). babies to the breast within 1 hr of delivery, and only a few (4.6% of intervention mother and 9.5% of the control mother) reported an intention to give prelacteal feeds to their babies (Table 2).

| Characteristics of delivery
The characteristics of delivery were similar in both groups. One half of mothers in both groups delivered their babies in health facilities (intervention 52.6% vs. control 51.8%; p = .985), whereas approximately one fourth of the mothers in both groups had home deliveries. More than half of the mothers had normal vaginal deliveries, and approximately 31% of both groups had caesarean sections (p = .346).
Forty-eight percent of the mothers in the intervention group and 43.7% in the control group were assisted by qualified doctors during delivery (p = .695), and over 20% of mothers in both groups were assisted by trained birth attendants (p = .681; Table 3).

| Effects of the intervention on EIBF and EBF
The infant feeding practices of the mothers in both the intervention and control groups are shown in Table 4. A significantly higher proportion of mothers in the intervention group reported EIBF compared with the control group (89.1% vs. 77.4%, p = .005). More than 95%  of mothers in both groups gave colostrum to their babies as their first decreased from 86% at 1 month, 81% at 3 months, and 73% at 5 months, whereas the corresponding rates in the control group were 71%, 49%, and 27%. The differences in the EBF practices between the intervention and control group at 5 months were significantly different (p < .001; Figure 2).

| Association of EIBF and EBF with background characteristics
In the adjusted multivariable model, the most significant determinants of a lack of EIBF across the two groups were the mother's education, caesarean section, and low birthweight.  (Tables 5 and 6).

| DISCUSSION
This study provides evidence that peer counselling positively influences EIBF and EBF practices among mothers living in an urban slum. The low prevalence of EBF in the control group and high prevalence of EBF in the intervention group confirm the need for community-based peer support to promote breastfeeding in urban slum areas. Peer counselling was found to be an effective strategy and led to much higher rates of EBF in the first 5 months compared with the control group.
In our study, we found that the rate of EIBF in the intervention group was 89.1%, which is higher than the values reported for studies in Pakistan and India, but lower than that in Nepal (Khanal, Scott, Lee, Karkee, & Binns, 2015;Patel et al., 2015;Zafar, Fatmi, & Shafi, 2014).
There are multiple factors, including maternal education, multiple births, type of delivery, not putting the baby to the breast after birth, and low birthweight, that are associated with a lack of EIBF (Acharya & Khanal, 2015;Gul, Khalil, Yousafzai, & Shoukat, 2014;Khadduri et al., 2008;Sharma & Byrne, 2016  Approximately one third of mothers in both groups delivered their babies via caesarean section, and these mothers were nine times less likely to initiate breastfeeding immediately after birth. This could be because the newborn baby is not usually put to the mother's breast    More than 60% of deliveries still occur at home in Bangladesh (Sarker et al., 2016), so our finding that approximately one third of women delivered their babies at home in both groups implies the likelihood of delivering in a health facility is higher for urban mothers. Our results indicate that mothers who had facility-based deliveries were less likely to initiate breastfeeding than mothers who delivered at home. This is in contrast to studies in Nepal and Nigeria, which found mothers who deliver in a health facility were more likely to initiate early breastfeeding compared with mothers who delivered at home (Adhikari, Khanal, Karkee, & Gavidia, 2014b;Berde & Yalcin, 2016).
The significantly lower proportion of neonates who were given any type of prelacteal feed in the intervention group implies that peer counsellors had a significant influence on reducing prelacteal feeds. Similar findings were reported in earlier studies from India (Kushwaha et al., 2014), Burkina Faso, Uganda, and South Africa (Tylleskar et al., 2011).
Regarding the rate of EBF in developing countries, global data suggest the prevalence of EBF among infants younger than 6 months increased from 33% in 1995 to 39% in 2010 (Haroon et al., 2013). Two previous RCTs have assessed the ability of community-based peer counselling to promote EBF, but neither followed up the mothers for as long as 5 months. Haider et al. (2010) were the first to explore the impact of peer counselling in Bangladesh and found that 70% of the mothers in the intervention group practiced EBF but only 6% in the control group practiced EBF at 5 months. A similar approach was used in this trial, and we found 73% of mothers in the intervention group and 27% in the control group practiced EBF at 5 months. The overall nationwide improvements in EBF practices, due to social campaigns, media coverage, programs run by national and international NGOs, and so forth, may possibly explain the higher rate of EBF in the control group in this study compared with that of Haider et al. (2010).
The participants in this study lived in urban slum communities, and the EBF rate of the control group was similar to the value of 16% reported in the recent study (Akhtar et al., 2012).
Thus, this study demonstrates one-to-one peer counselling may be an effective strategy to improve EIBF and EBF practices in urban slum communities, where a high rate of malnutrition exists and inappropriate breastfeeding practices are prominent (Shakya et al., 2017). Our peer counsellors delivered a regular service on a part-time (half-day) basis and received a monthly honorarium of approximately US $43.
The counsellors also made additional visits if necessary and conducted fortnightly group sessions with the mothers and other family members. In a lower resource setting, fewer visits combined with Note. aOR = adjusted odds ratio; OR = odds ratio.
group counselling may be effective to increase the rates of EIBF and EBF (Haider et al., 2000). Governments and NGOs could even consider training community health volunteers to provide group or peer support on infant and young child feeding practices to the mothers and caregivers.
One strength of this study is its randomized design, which minimized selection bias (Sibbald & Roland, 1998). Moreover, the power of randomization provides excellent internal validity, which is one of the greatest strengths of RCTs. Randomization ensures that the level of exposure to the treatment of interest is the only differentiating factor between the two arms. Strength of this study is that the intervention combined individual counselling with fortnightly group sessions. Including pregnant and lactating mothers and family members in counselling and practical demonstrations has also been proven to be effective (Sibbald & Roland, 1998).
One limitation of this randomized study is that it was not double blinded, and the interviewer knew which clusters were included in the intervention group. To prevent interviewer bias, the questions related to peer counsellor contact were asked to the mothers at the end of each follow-up interview session. Additionally, the generalizability (external validity) of this study may be limited (Meyer, 2010), as there was a high dropout rate during the trial (25% in the intervention group and 16% in the control group). In most urban slums in Bangladesh, including Mirpur, people often move from one slum to another in search of better employment, cheaper accommodation, or in response to political unrest. Moreover, many working mothers who formerly resided in urban slums often quit their jobs to take care of their children and migrate back to rural areas. This phenomenon could explain the high dropout rate and loss to follow-up in this study.
In summary, we found that peer counselling was an effective method of promoting EIBF as well as EBF among urban women living in a slum in Dhaka, Bangladesh. This result is very promising, as this slum community has exceedingly low rates of EBF and a strong preference for formula, even if mothers choose to breastfeed their infants (Chapman et al., 2004;Perez-Escamilla et al., 1998). We have shown the potential of peer counselling to promote EIBF and EBF among women residing in urban slum areas.

| CONCLUSION
Peer counselling positively influenced the initiation of breastfeeding within 1 hr of birth and the duration of EBF and reduced prelacteal feeding among urban women living in a slum in Dhaka, Bangladesh. This study highlights the need to target the promotion of optimal breastfeeding practices to expectant mothers and mothers who deliver through caesarean section.
Furthermore, peer counselling support to support EBF should be continued for at least first months of delivery.

AVAILABILITY OF DATA AND MATERIALS
The dataset has been uploaded as Additional File 1 (EBIF.dta).

ACKNOWLEDGMENTS
This study was conducted at the icddr,b Centre for Health and