Breastfeeding is a well established and recommended intervention for the improvement of child nutrition. Studies have demonstrated that it reduces deaths in infants and young children [1–3]. It is one of the most important factors for growth and development of infants and is globally endorsed as being the best for any neonate [4]. The World Health Organization (WHO) recommends the practice of exclusive breastfeeding (EBF) of infants for the first six months of life after birth. EBF means that the infant receives only breast milk. No other liquids or solids are given, not even water, with the exception of oral rehydration salt solution, or drops/syrups of vitamins, minerals or medicines [5].

Throughout the world, in developed and developing countries alike, inappropriate feeding of infants leading to their poor nutrition is a significant problem affecting socio-economic progress in general [6]. Suboptimum breastfeeding was responsible for 11.6% of all child deaths in 2011 [7]. Universal (90%) coverage of breastfeeding is estimated to prevent around 13% of all deaths among children under five years of age in low and middle income countries [3]. However, globally only half of infants under 1 month of age and 30% of infants aged 1–5 months are exclusively breastfed [7].

Previous studies have reported several predictors of EBF. Differences are evident not just between countries but also within the same country. Several factors have been shown to be associated with EBF: variations between urban and rural areas, infant’s age, mothers’ employment status and education level, knowledge about good breastfeeding practices, occupation, monthly household income, mothers’ smoking status, socio-economic position, prelacteal feeding, parity, positive attitudes towards EBF, intent to exclusively breastfeed before delivery, timely initiation of breastfeeding, mode of delivery, infant’s birth weight, health system practices, discarding colostrum and community beliefs [8–14].

In Bangladesh, the prevalence of EBF has remained largely unchanged for nearly two decades. It was around 45% in 1993–94 and 1999–2000 [15, 16], then declined to 42% in 2004 [17] and was 43% in 2007 [18]. However, in 2011, a prevalence of 64% was reported, an increase of 21 percentage points [19]. The reasons for this large change remain speculative at this point. Thus to investigate the issue further, this study was conducted in a rural sub district where EBF prevalence had not been investigated by any prior studies. The objective was to assess the prevalence of EBF and associated factors among mothers having infants aged 0–6 months in rural Bangladesh.

A cross-sectional study was carried out in Mirzapur Upazilla (Sub district) of Tangail district in rural Bangladesh among mothers having infants aged 0–6 months. The rationale for selecting the site was: Tangail is ranked 37^th for the timely initiation of breastfeeding out of 64 districts, and Mirzapur sub-district under Tangail district is ranked 79^th for timely initiation of breastfeeding out of 483 sub-districts of Bangladesh [20]. Thus, Mirzapur is a typical sub-district of Bangladesh. It is neither a ‘high performing sub-district’ nor a ‘low performing sub-district’. In addition, there were no focused nutrition interventions including promotion of breastfeeding programmes going on in that area. A baseline prevalence of 43% [18], margin of error 10%, with 95% CI was used to calculate the sample size. Further a design of effect 1.2 was added [21] which resulted in a sample size of 121. Taking into account a non-response of 25% the final sample size was calculated to be 151.

The list of mothers having infants aged 0–6 months was obtained from the demographic surveillance systems (DSS) computerized database that is updated on a weekly basis by field staffs of International Centre for Diarrhoeal Disease Research, Bangladesh (icddr, b). The list comprised of 1206 mothers having infants aged 0–6 months. From that list, required number of mothers in each stratum (0–2 months, 3–4 months and 5–6 months) was proportionately calculated and randomly selected using SPSS (version 20) software (Figure 1). A total of 151 mothers were randomly selected. The households were then identified and mapped for ease of data collection. Out of 151 mothers, 30 were absent during interview and were not followed up thereafter. In the end, we interviewed and analyzed data from 121 mothers: 12 mothers of 0–2 month old children, 47 mothers of 3–4 month old children and 62 mothers of 5–6 month old children. Data were collected in November 2012 by using pre-tested semi-structured questionnaire that inquired information on socio-demographic characteristics, obstetric, health service and breastfeeding related factors (initiation of breastfeeding, prelacteal feeding and colostrum feeding) and economic factors. Questions related to EBF were asked using 24 hour recall method. Survey questionnaires were administered in native language (Bangla) of the respondents.

Data were entered in SPSS (version 20) and analyzed using SPSS (version 20) and Stata (version 12). The outcome variable was EBF as per WHO definition [5]. Wealth index was constructed using household asset data [18]. It comprised of ownership of several consumer items. Principal component analysis was used to generate and assign weight (factor score) to each asset. The asset scores for individual household were then computed and divided into quintiles from one (poorest) to five (richest). Univariate analysis was carried out for all variables to describe their characteristics. In bivariate analysis, crude odds ratio (OR) with 95% CI and Chi-square test or Fisher’s exact test (as appropriate) were calculated to identify associations between categorical variables. In multivariate analysis, a logistic regression model was developed to identify the associations between the dichotomous outcome variable: EBF (yes/no) and potential predictors. The final model was generated by stepwise backward elimination procedure and adjusted odds ratios (AOR) were calculated with 95% CI. Thus, the final model had only those predictors that showed some evidence of association with the outcome variable.

Ethical approval was obtained from the ethical review committee at James P. Grant School of Public Health, BRAC University, Dhaka. Prior to interview, written informed consent was obtained from the mothers. In cases where the mother was illiterate, verbal consent was taken from the mother following a detailed description of the study protocol. The interviewer also signed a statement for verifying that he had provided the necessary information written in the informed consent form and that each respondent had consented to participate in the study. All data were coded to remove identifying information and secure confidentiality. Respondents were not compensated for participation in the study.

EBF is the best recommended infant feeding method for the first six months of life and has a protective effect against child morbidity and mortality [22]. But it has not yet been universally practiced and the reduction in the rate of EBF is taken as a serious problem, especially in developing countries [22]. In the present study, prevalence of EBF was lower (36%), compared to national level (64%) [19]. The Bangladesh Demographic and Health Survey (BDHS) results were from the whole country which includes nutrition intervention and non-intervention districts. In comparison, our study was conducted in a non-intervention district which did not have any nutrition and breastfeeding promotion programmes. This might be one of the reasons why prevalence of EBF was lower in our study. The BDHS has attributed the higher prevalence of 64% to the inclusion of greater number of young infants in their sample [19]. Mothers are more inclined to exclusively breastfeed their children when they are young and this could result in over estimation of EBF prevalence. In contrast, the number of young infants (0–2 months) in our study was too low. As a result we could not make direct comparisons with the BDHS result. Thus we have disaggregated the prevalence by age and reported for older age groups as well. Additionally, other reasons for low prevalence in our study may include lack of support from family members, lack of advice from health staffs during ANC visits and deliveries in non baby friendly institutions. The baby friendly hospital initiative was established by WHO and UNICEF in 1991 as a hospital-based intervention to increase breastfeeding rates. In baby friendly hospitals breastfeeding is supported, practiced, protected and promoted [23].

Chudasama et al. reported prevalence of EBF to be 37% in Gujrat, India [24]. Another study from India among rural women in Tamil Nadu showed the prevalence to be 34% [25] which is similar to our study. Results may be similar because both the study sites were rural and both countries have similar type of cultural practices. Statistically significant differences were not observed between those who exclusively breastfed and those who do not with respect to variables that are considered to be supportive for breastfeeding such as maternal education, number of ANC visits, breastfeeding support and education, and place of delivery which was similar to a study conducted in India [26].

Children are more likely to receive prelacteal feeds when they are born at home and when the birth is not assisted by a health professional [19]. Over the years there has been a decline in the percentage of prelacteal feed at the national level - 62% in 2007 [18] and 39% in 2011 [19]. This could be explained by the decrease in the percentage of home deliveries - 85% in 2007 [18] and 71% in 2011 [19]. This might as well apply to our study where prelacteal feeding and home deliveries both are comparatively low.

Even though the study site was a fairly typical low-income upazilla, the presence of a large charity hospital-"Kumudini Hospital" meant that services were provided free of cost to all those who were unable to afford medical care; as a result of which a large number of residents sought health care services from the hospital. This might be the reason why many deliveries were institutional and assisted by doctors or nurses. On the contrary, at the national level 71% of the births were delivered at home [19].

Type of delivery was associated with EBF status. The mothers who delivered by caesarean section had lower odds of EBF than those who delivered normally. Ideally, all hospitals are required to practice ten steps to successful breastfeeding recommended by WHO [27] but it is seldom practiced. This could be one of the reasons why mothers who delivered by caesarean section did not practice EBF. As a result, the baby may be fed prelacteals even before breastfeeding is initiated for the first time. Furthermore, the mother and her family often think that breast milk itself is not sufficient and infant formula is necessary which is easily available around the hospital premises. A study from China [28] has reported that after surgery (c-section) mothers feel pain around the incision area and there is difficulty in movement because of catheterization and intravenous lines. The authors of this study argued that this could be a reason for non EBF. Similar explanation could also be valid for our study. Researchers from Nepal have also reported low EBF rates among mothers who delivered by caesarean section [29]. Further studies are needed to better understand the influence of caesarean section on EBF.

We found that the mothers from households belonging to richer wealth quintile had higher odds of EBF than those belonging to poorest wealth quintile. In contrast, other studies [17, 30] have reported prevalence of EBF to be higher among children belonging to poorest wealth quintile. Mothers belonging to richer wealth quintile may have better education level, easier access to media and health services which may have increased their awareness and made them relatively more conscious about EBF.

There may be an intrinsic relation between EBF and its determinants but the present study did not aim to discuss those in detail. The present study alerts us that even though there is no relation with its determinants, the prevalence of EBF is very low which needs to be addressed. Further studies are necessary to investigate several other factors which are known to have relation with EBF. In addition, other psychological factors which include emotional symptoms, peer problems, or social behavior, parent's marriage or the mother's satisfaction with her relationship with her partner or child may be explored. Anthropological questions including the influence of conflict, work place/condition, culture, tradition, and environment need to be answered.

In our study, there was a non-response of 25%. Several factors could be held accountable for it. First, the respondents who were not available had recently migrated from the study site. Second, there was a culture of visiting maternal home (respondent's parents’ home) soon after the birth of the child. Third, either the respondent or her child was hospitalized due to some illness. Further limitations of this study include the use of 24-hour recall data for the calculation of EBF which may be inadequate. Despite the limitation, the strength of the study was the use of unbiased sample size with stratification, appropriate sampling design and analysis plan and selection of a study area where there was no nutrition intervention which resulted in unbiased estimates. If the study would have been conducted in a nutrition intervention area then the information collected would have been greatly influenced by the programme activities. An additional strength of our study is that it was conducted in a DSS site of icddr,b where the use of a computerized database of households with eligible respondents for sampling helped to increase the internal validity. The database was well maintained and updated weekly.

The prevalence of EBF (36%) in Mirzapur is lower than the national figure (64%). Prelacteal feeding was not uncommon. These findings indicate that the breastfeeding support provided by health services is weak. Further, there were no nutrition interventions and breastfeeding promotional activities going on in that area. Hence, promotion of EBF during the first six months of life and continuation of breastfeeding after six months along with appropriate complementary feeding, and focus on the factors associated with EBF need to be addressed. Future breastfeeding promotion programmes by the government and partners should give special attention to those women who are not practicing EBF.