The effect of mother’s educational status on early initiation of breastfeeding: further analysis of three consecutive Nepal Demographic and Health Surveys

Background The World Health Organization recommends initiating breastfeeding within the first hour of birth. This study is aimed at assessing the effect of the mother’s education on early initiation of breastfeeding. Methods Data from the Nepal Demographic and Health Surveys (NDHS) 2001, 2006 and 2011 were used which included 12,845 last born children born within 5 years before the surveys. Early initiation of breastfeeding was defined as the initiation of breastfeeding within the first hour after birth. Hierarchical modelling was used to ascertain the association of maternal education and early initiation of breastfeeding, after controlling for other covariates in a multiple logistic regression. Results Maternal education was associated with a higher likelihood of early initiation of breastfeeding in each survey. Pooled data analysis revealed higher odds of early initiation of breastfeeding among the mothers with primary education (adjusted odds ratio (OR) 1.24, 95 % confidence interval (CI): 1.09, 1.42) and secondary or higher education (OR: 1.63 95 % CI: 1.42, 1.88). In the most recent NDHS 2011 survey, odds of early initiation of breastfeeding was higher among mothers with primary education (OR: 1.52; 95 % CI: 1.21, 1.91) and mothers with secondary or higher education (OR: 2.20; 95 % CI: 1.76, 2.76) compared to mothers with no education. Similarly, the odds of early initiation of breastfeeding was higher among mothers with secondary and higher education in the 2006 data (OR: 1.66; 95 % CI: 1.30, 2.12) and in 2001 (OR = 1.30; 95 % CI: 1.00, 1.67). Conclusions As the association between a mother’s educational status and her likelihood of early initiation of breastfeeding increases, long-term approaches to prioritising education for women and girls should be explored. In the short term, uneducated mothers should be targeted with breastfeeding promotion strategies such as counselling and peer education. Electronic supplementary material The online version of this article (doi:10.1186/s12889-015-2405-y) contains supplementary material, which is available to authorized users.


Background
Globally, about 60 % of child deaths are directly or indirectly linked to undernutrition. Two-thirds of child deaths are estimated to be attributed to improper feeding during the first year of life [1]. As the leading global health institution, the World Health Organization (WHO) has identified sub-optimal breastfeeding as a barrier to sustainable socioeconomic development and poverty reduction [1] and recommends initiating breastfeeding within the first hour after birth [2][3][4]. The first breast milk after delivery (colostrum) provides energy and immunity to newborn [5]. Delayed initiation of breastfeeding may lead to hypothermia [6]. A study has reported higher odds of neonatal mortality [pooled odds ratio (OR): 2.02, 95 % confidence interval (CI): 1.40, 2.93] among those who were breastfed after the first hour of birth compared to the children breastfed within the first hour after birth [7]. Similarly, a hospital based cross sectional study in Ghana reported an increased risk of infant mortality in the case of delayed initiation of breastfeeding. The study reported greater than a twofold risk (OR: 2.4, 95 % CI: 1.69, 3.40) of mortality if the initiation was delayed for 1 day after birth [8]. In Nepal, the risk of infant mortality was found higher among those who were breastfed later than 24 h of birth (relative risk (RR): 1.41; 95 % CI: 1.08, 1.86) [9].
Nepal has achieved the target for the millennium development goal(MDG) target on the reduction of child mortality. However, neonatal mortality is still alarming at the rate of 33 per 1000 live births and was relatively stagnant in the past decade [10]. A cross-sectional study from central Nepal in 2005 reported that 72.7 % of recently delivered women initiated breastfeeding within the first hour of birth [11]. More recently, a community based prospective cohort study from central Nepal showed that about 67 % mothers initiated breastfeeding within first hour after delivery [12]. A further analysis of the NDHS 2011 reported that about 27 % of mothers provided prelacteal foods to their newborn babies [13]. That study found maternal education associated with the prelacteal feeding. Similarly, studies from abroad have found a positive association between maternal education status with the early initiation of breastfeeding [14,15], exclusive breastfeeding practice [16][17][18] and the duration of breastfeeding [19]. However, early initiation of breastfeeding and the factors associated with early initiation remains understudied in Nepal. Education status of the mother has been identified as an important social determinant of health for children. The role of maternal education on infant feeding behaviour is an interesting area to explore. Women's education in Nepal has long been neglected and still remains much lower than men. In recent years, there has been an improvement in women's education status. Therefore, the effect of maternal education might also impart effect on health and infant feeding behaviour. This study aims to investigate the association between early initiation of breastfeeding and mother's educational status using data from the Nepal Demographic and Health Survey 2001, 2006 and 2011.

Methods
The Nepal Demographic and Health Survey (NDHS) is a nationally representative population based survey conducted every 5 years since 1986. The NDHS implemented two stage cluster sampling. Firstly, the country was divided into sample domains and secondly sampling units were selected in each sample domain with probability proportional to size. The household was selected randomly from the sampling unit. The overall response rates of these surveys were high: NDHS 2001 (97.8 %), NDHS 2006 (98.0 %) and NDHS 2011 (97.6 %). Details of the sampling procedure are mentioned in the individual survey reports [20][21][22]. Child datasets of the NDHS 2001, 2006 and 2011 were  appended for analysis. The datasets contained information about under 5 year old children, their parent and household characteristics. A total of 12,845 last born children born within 5 years before individual surveys were included in this study irrespective of their current status (living or dead). Since the time of initiation of breastfeeding is applicable to only those children who were reported to be breastfed at least once, children who were never breastfed were excluded from analysis.

Definitions of variables
WHO defines early initiation of breastfeeding as 'provision of mother's breast milk to infants within one hour of birth' [23]. During the survey mothers were asked -"How long after birth did you first put (name of the children) to the breast?"The answers were recorded in terms of hours and days after birth. For this analysis, a child breastfed immediately after the birth and within first hour of birth was coded as "1 (Early initiation)" otherwise"0 (Late initiation)" [1,4].
The main explanatory variable was mother's education level which was recorded as: (i) no education (ii) primary education and (iii) secondary or higher education. Other independent variables were selected based on literature [19,24]. Place of residence was reported as (i) urban and (ii) rural. Ecological zone was classified according to the altitude as (i) Mountain (ii) Hill (iii) Terai. Administrative regions are the five north-south segments of the nation namely (i) eastern development region (ii) central development region (iii) western development region (iv) mid-western development region and (v) far-western development region. Mother's occupation was classified into 3 categories: (i) not working (ii) agriculture (iii) other paid jobs [24]. Father's education was recorded similar to mother's education. Religion was grouped into (i) Hindu and (ii) Others (Muslim, Buddhist, Christian). Ethnicity was firstly classified in to seven groups according to Nepal Demographic and Health Survey ethnicity classification [25] and was further grouped into four categories:(i) relatively advantaged (Brahmin/Chhetri) (ii) relatively disadvantaged (Janajati/Newar/Muslims) (iii) Disadvantaged (Dalits) and (iv) others (Madheshi and other unidentified) [25]. NDHS uses household assets to derive wealth index derived from principal component analysis [26] as a measure of socioeconomic status. This wealth score is then divided in to five quintiles (i) poorest (ii) poorer (iii) middle (iv) richer and (v) richest [22]. The mother's age was grouped into three categories (i) < 20 year (ii) 25-34 years (iii) > =35 years. The sex of the child was grouped as: (i) male (ii) female. Birth order was recoded as: (i) first (ii) second (iii) third or more. Size at birth was based on the mother's perception of the size of child at birth which was recorded as:(i) very small/small (ii) average or (iii) large/very large [13]. Variable multiple births is a dummy for twins/triplets. Place of delivery was classified as: (i) home (ii) health facility (government or private facilities) (iii) other places. Mode of delivery was recorded as: (i) caesarean (ii) vaginal.
The DHS are conducted in more than 50 countries, mostly in low and middle income countries, and serve as a major source of policy making in those contexts. A generic question was first prepared by measure DHSprogram and modified according to country specific context. In Nepal, the questionnaire was pre-tested and validated before the survey. More detail on the questionnaire can be found elsewhere [20][21][22]27].

Statistical analysis
We adopted the conceptual framework from Matanda et al. [28] with some modifications (see Fig. 1) to facilitate our statistical analyses. This conceptual framework was originally based on the model developed by the United Nations Children's Fund (UNICEF) and extended by Engle et al. [29]. The variables were grouped in to context, socio-economic and child specific demographic/health care variables according to their proximity to the dependent variable. We assumed that proximate variables have more direct effect than the distant variables. Statistical analysis was based on the survey analysis technique adjusting for weights assigned for non-response (sample weight) and cluster sampling design. Descriptive analysis was performed for sample background characteristics, followed by a logistic regression analysis to assess the association between mother's education status and early initiation of breastfeeding, based on the conceptual framework. Context variables (urban/rural, ecological zone and administrative regions) were adjusted in model-1. In model-2; socioeconomic variables (mother's occupation, fathers' education status, religion, ethnicity, wealth quintile) and in model-3 child specific demographic and health care related variables (mother's age, ANC utilization, sex of child, birth order of the child, the size of the child at birth, multiple birth, place of delivery, mode of delivery) were adjusted respectively. Variables found significant in the previous model (P < 0.05) were additionally adjusted in succeeding models until the final model (model-4) was obtained with all the significant variables by following the conceptual framework [30].

Mother's education Mother's occupation Father's education
Mother's age Sex of child Fig. 1 Conceptual framework of determinants of early initiation of breastfeeding, prepared for hierarchical logistic regression modeling. Based on the model developed by UNICEF and extended by Engle et al [28]. Adopted and from Matanda et al [29] with some modification We checked the interaction term 'survey year*mother's education' to assess the change in mother's education status between the surveys, which was significant at p < 0.001. Therefore, independent analyses were also performed for each survey 2001, 2006 and 2011. P-value < 0.05 was considered statistically significant. Statistical analysis was performed in STATA 13.0.
Ethics NDHS obtained ethical approval from the Nepal Health Research Council, Kathmandu, Nepal and Macro Institutional Review Board, Maryland, USA [20][21][22]. NDHS obtained consent from mothers and their children before starting the interview. Permission to use the survey data for this study was granted by the Measure DHS program [27].

Characteristics of respondents
The result represents a total of 12,845 mother-child pairs from demographic and health surveys. The majority of the respondents (58.68 %) had no education, about 17.50 % had primary and 23.82 % had a secondary or higher level education (see Fig. 2). Nearly nine in every ten respondents were from rural areas, more than half (51.01 %) were from Terai/plain land and about 31.93 % were from the central development region. About three quarters (75.33 %) of mothers were working in the agriculture sector. The majority of the children (61.29 %) were born from the mothers of age group 20-34 years. Only about 21.97 % children were delivered in a health facility ( Table 1).

Percentage of children breastfed within the first hour after birth
The pooled data showed that slightly more than one third (36. Fig. 3). The difference in the percentage of women initiating breastfeeding within first after birth was statistically significant (P < 0.001) during the survey periods. The proportion of early initiation from pooled estimation was found to increase with increase in level of education. Results from the pooled data showed that children from the mothers with at least secondary education were more likely to be breastfed within 1 h after birth (48.29 %) in comparison to mothers with no education (31.55 %) and mothers with primary education (38.54 %) (see Fig. 3). A similar pattern was found, according to the father's education status.
The percentage of newborn infants breastfed within the first hour after birth was higher among urban (41.44 %) areas in comparison to rural areas and the hill region (40.29 %) compared to the mountain or Terai region. Similarly, babies from the mothers with paid jobs, whose fathers had higher education, from the Hindu religion, from the advantaged ethnic group; from higher wealth quintiles had higher rates of early initiation of breastfeeding. The percentage of early breastfeeding was higher among the babies from mothers of age 20-34 years, mothers who completed four antenatal care visits, babies born in the second in order, babies average or larger in size at birth and babies delivered in a health facility. The proportion of early breastfed babies was lower among the multiple  births and caesarean delivery compared to babies delivered by vaginal delivery. Detail comparison of proportions is presented in Table 1. These models showed that there was increased likelihood of early initiation of breastfeeding with increasing maternal education. This analysis found other characteristics independently associated with early initiation of breastfeeding, after controlling for the effect of mother's education. The groups who had a higher likelihood of early initiation were the mothers who were from mid-western and farwestern regions, from higher wealth quintile, who had four antenatal care visits, were second parity mothers, had larger babies, and had delivered in health facilities. Conversely, mothers who were from the Terai (plain) region, rural areas, from madheshi and other ethnic group, had multiple births and had delivered via caesarean section had a lower likelihood of early initiation of breastfeeding. (Details are shown in the Additional file 1).

Discussion
This study found newborn babies from the mothers with primary education and secondary or higher education were more likely to be breastfed within the first hour after birth in comparison to the babies from the mothers with no education. We found a significant increase in the percentage of early initiation of breastfeeding during three surveys. National nutrition policy and strategy 2004 and its amendment in 2008 have given special emphasis on breastfeeding promotion including early initiation of breastfeeding [31], also mentioned in the Multi-Sectoral Nutrition Plan (2013-2017) [32]. Community based newborn care program (CB-NCP) implemented by the government directly advocates for the initiation of breastfeeding within first hour after birth [33,34].
The government has promoted Baby Friendly Hospital Initiative [35][36][37]. Public awareness campaigns via mass media might also have impacted to this increase in early initiation of breastfeeding. A recent study conducted in central Nepal reported that about 67 % of recently delivered mothers breastfed their newborns within the first hour of birth [12]. A cross sectional survey conducted in an urban city of western Nepal in 2006 found that about 58 % mothers initiated breastfeeding their babies within first hour after birth [38]. These studies reported higher rates of early initiation than the NDHS and the difference can be accounted for in varied study settings and methodology. This study confirms the positive role of mother's education to early initiation of breastfeeding. The association between mother's education and early initiation of breastfeeding was at a marginal significance level for mothers with secondary or higher education in 2001 and was statistically significant in 2006. Meanwhile, the association was more prominent for the latest 2011 survey. This shows the increasing importance of education for uptake of 'messages' about breastfeeding that emerged in government policy and various national programmes in between the survey periods. An observational study from Bangladesh reported that caretakers with formal education were more likely to feed their babies more frequently, feed hygienic foods in a hygienic environment [39]. A review of literature had reported maternal education as a consistently associated demographic factor with early initiation of breastfeeding [19]. A systematic review of interventions designed to promote exclusive breastfeeding in high income countries found that the interventions using educational approaches were significantly associated to the increase in the duration of exclusive breastfeeding [40,41]. Educating and counselling mothers have found as a successful   intervention for early initiation and sustaining breastfeeding even in the complicated situation like caesarean section [42]. Taken together, the current findings of a consistent association of maternal education with early initiation of breastfeeding is comparable to international findings.

Other significant factors associated with early initiation of breastfeeding
It is worthwhile to discuss the other demographic and socio-economic characteristics identified as independently associated with the early initiation of breastfeeding. However, the primary objective of this paper was to investigate the association between mother's education and early initiation of breastfeeding. Higher odds of early initiation of breastfeeding in the western and far western region can be explained by the concentrated effort of existing maternal and child health intervention programs in the recent years. It might also because of the unavailability or inaccessibility to the alternatives of breastfeeding in those regions.
The odds of early initiation were higher among higher wealth quintiles. Socio-economic deprivation has been identified as the predictor of breastfeeding duration elsewhere [43,44]. Higher wealth index was found positively associated with the greater likelihood of using recommended level of antenatal care visits [45], where they receive counselling about colostrum feeding. Wealthier mothers are also more likely to deliver at the health facility and receive postnatal care where they can get more support for breastfeeding initiation [46,47], while poorest mothers are deprived from those advantages.
Higher odds of early initiation of breastfeeding among the babies delivered at health facilities can be explained by the presence of health professional, their monitoring and support for the breastfeeding, which assures mothers to start early initiation of breastfeeding. This study found a significant proportion of children delivered at health facility were still not breastfed within first hour after birth. A study reported found higher rate of early initiation of breastfeeding and better newborn care practices among the institutional delivery in Dhanusha district (from the Terai region) of Nepal [48]. However, the study also showed the scope for improvement on the early initiation of breastfeeding among institutional deliveries [48].
Similarly, large size babies are more likely to be full termed and healthier, they have better coordination of the suction-deglutition-respiration cycle and breast seeking reflex [49,50]. This explains why the large size babies are more likely to be early breastfed than small size babies. Support from the family and society is important factor for successful breastfeeding. Women's knowledge about the benefit of breastfeeding and their perception plays an important role to initiate breastfeeding early. Women in rural areas are found to have low knowledge about the importance of breastfeeding and they are also less likely to get support for breastfeeding. Those factors might have resulted in lower odds of early initiation of breastfeeding in the rural areas [51]. Similarly, lower odds of early initiation of breastfeeding in the terai region may be associated with the cultural practices of introduction of prelacteal food in the region [13]. An earlier study from Rupandehi district of Nepal (from the terai region) also reported extremely low rate (3.4 %) of initiation of breastfeeding within the first hour after birth [9].
A study in southern Nepal, where most of the participants were from Madheshi ethnic group revealed that about two in five mothers discarded colostrum [52]. Such harmful cultural practices inhibit initiation of breastfeeding within first hour after birth; therefore we obtained lower odds of early initiation of breastfeeding among those ethnic groups.
It may be difficult for mothers to breastfeed early who are primi-parous, have multiple births and have a caesarean delivery to initiate breastfeeding within the first hour of birth. This may explain the lower odds of early initiation among caesarean deliveries [53][54][55][56]. Earlier studies have found parity associated with the breastfeeding behaviour [57,58]. Confidence of mother is also an important factors which determines the initiation of breastfeeding, which increases with the parity [56].

Public health implications of the current findings
The current findings have a number of public health implications. This study has demonstrated the importance of maternal education on early initiation of breastfeeding. Since, improving educational status of mothers is a long term agenda, immediate attention should be provided to mothers with no or less education with alternative supportive and educational interventions like prenatal education and counselling [59]. Midwives, other health professionals and nurses can play important role to encourage mothers for early initiation of breastfeeding [60].
This study is based on the three large, nationally representative population based Demographic and Health Surveys in Nepal. Large sample size ensures a high precision of the findings. NDHS used the standardized tools; therefore, the results are reliable and comparable with other developing countries. However, causal inference between maternal education level and early initiation of breastfeeding is limited due to the study being cross sectional. Similarly, maternal education here represents the formal education gained through schooling and it may not reflect the health literacy of the mothers.