1Child mortality under five years of age has drastically declined in Niger in the last decades from 328‰ in 1990 to 96‰ in 2015 whereas decrease in infant death in the first 12 months of life was slower shifting from 55‰ to 27‰ in the same period (UNICEF, 2016). In 2013, the new Lancet series on maternal and child nutrition delivered a clear message on risks of suboptimal breastfeeding in the first 2 years of life which lead to an estimated 800,000 child deaths per year (Black et al., 2013). Many highlighted urgent actions to focus on reducing newborn deaths (Bhutta et al., 2013), among others optimal postpartum breastfeeding by putting a child to the breast within the first hour of birth which is still an uncommon practice in rural Niger. Half of mothers in Niger initiate breastfeeding within the first hour after delivery (52.9 %), and those who breastfeed their child exclusively until the age of 6 months are even fewer (23.3 %) (Institut National de la Statistique - Niger, 2012). The benefit of early initiation of breastfeeding remains insufficiently exploited (Darmstadt et al., 2005).

2The bottleneck for breastfeeding promotion is that mothers encounter a number of constraints in their family and community in Niger. Behavior change to promote “optimal” postpartum breastfeeding should lead to creating social, environmental and physiological conditions to encourage mothers to breastfeed their child in line with WHO/UNICEF’s recommendations (WHO, 2003). The experts consultation team delivered recommendations in a recent The Lancet series on child survival (2013) to outline the definition of “optimal” breastfeeding patterns known to be the most efficient and appropriate practices for postpartum, infancy and child survival under three categories: early initiation of breastfeeding within the first hour of birth, exclusive breastfeeding until 6 months and continued breastfeeding until 24 months.

3While early initiation of breastfeeding is an act of putting a child to breast immediately after birth, it also implies exclusivity, meaning that mothers do not give their child anything but breast milk. The cut-off point set by WHO is within one hour of birth (WHO, 2003) to be defined as “early initiation” of breastfeeding, and any practice beyond the first hour after birth is defined “delayed initiation of breastfeeding”. Early and exclusive initiation of breastfeeding is recognized to be an indispensable practice to neonatal survival and health. Previous studies showed that initiation of breastfeeding within 24 hours of birth reduces neonatal mortality by 45 % (Bhutta et al. 2013) and combined with Kangaroo Mother Care (KMC), provides a substantial benefit to postpartum breastfeeding practices through skin-to-skin contact (Moore et al., 2012).

4The shortage of research findings on breastfeeding promotion in Nutrition and Food sciences databases (Morris et al., 2008), especially in Sub-Saharan African countries, constitutes a real bottleneck to practitioners seeking to apply evidence from successful field interventions. Impact assessment of kangaroo mother care to improve neonatal survival is one of the top priorities in the area of research on community based health system strengthening as highlighted by the Lancet neonatal survival steering team (Yoshida et al., 2014). A further reduction of post-neonatal mortality needs to be addressed with integrated interventions combining family and community based approaches (Martines et al., 2005; Chopra et al., 2012). To achieve the Millennium Development Goal for child survival (MDG-4), research should focus on testing new participatory approaches to marginalized populations not only from the viewpoints of health service providers but also from that of community and family members.

5UNICEF launched participatory action research in collaboration with local NGOs in Niger since 2008 to promote Key Family Practices (hereafter referred to as KFP), optimal infant feeding and child health care practices within family and community settings. The Ministry of Health in Niger endorsed a priority strategic plan of action with UNICEF to promote 8 Key Family Practices that include: (1) exclusive breastfeeding until 6 months of age; (2) introduction of complementary food from 7 months of age; (3) washing hands with soap at critical moments; (4) sleeping under insecticide treated nets; (5) management of diarrhea with oral rehydration salts (ORS); (6) seeking health care to prevent child illnesses; (7) detection of signs of illnesses and referring to a health center; and (8) birth spacing intervals of at least 24 months. The program was first piloted in two regions of Niger, Maradi and Zinder (Institut National de la Statistique, Niger, 2012). A number of different behavior change communication strategies were applied through different channels of communication identified in a participatory manner. An integrated package of communication actions encouraged people to get actively involved in decision making processes. As a result, expected outcomes of communication interventions are not only their improved knowledge of infant feeding and child health care, but also people’s ownership of change (Waisbord, 2001, UNICEF, 2008). This participatory approach contributes to creating dynamic interactions between individuals and communities to make behavior change sustainable.

6An evaluation study was designed and conducted within program settings to measure the impact of the behavior change communication program interventions on infant feeding and child health care at family and community levels. This chapter presents part of the PhD research that the author accomplished in 2015 and aims to identify determinants of early initiation of breastfeeding related to marital status of mothers. A preliminary research finding implied that polygynous unions, the predominant marital status especially in rural areas, are more inclined to early initiation of breastfeeding than those in monogamous unions (Horii et al., 2014). In this chapter, determinant analysis verifies this outcome and develops the assumption that there must be influencing factors other than marital status of mothers in polygynous unions leading them to practice early initiation of breastfeeding.

7First we identify what characterizes those married in polygynous unions in terms of socio-economic and demographic status. Second we further examine to what extent determinants of early breastfeeding identified in the previous article (Horii et al., 2017) interact with the association between marital status and early initiation of breastfeeding.

Methods

8This is a secondary analysis of a quantitative cross-sectional survey designed and conducted in 2010 to evaluate the effects of the behavior change communication program on infant feeding and child health care in rural Niger. This survey (hereafter referred to as “Key Family Practices (KFP) survey”) compared retrospectively a group exposed to the program with an unexposed group.

9The villages were purposively selected for the intervention of the pilot program activities which initially begun in two regions: Maradi and Zinder, based on a few criteria, e.g. access to functional health huts and infrastructure facilitating community outreach. The sample of the KFP survey includes a post-hoc constitution of two groups of interviewed mothers: (1) the group residing in the villages exposed to the program in Maradi and Zinder; and (2) the group not exposed to the program at the time of data collection in Maradi, Zinder, Tillabéry and Tahoua. The exposed group included all villages where women groups, local associations and institutions were actively involved in promoting Key Family Practices favoring mother and child health care. The unexposed group was drawn from villages to be targeted in a future intervention area. The current program was extended to villages that were initially in the unexposed group when the data collection of this survey was completed in 2010.

10The study population of this secondary analysis was built based on a study sample extracted from the original dataset. They include female household members aged 15-49 years having at least a child of 0 to 23 months, currently pregnant or not. The total number of the eligible study population after cleaning the merged dataset was 2,091. For this secondary analysis, three surveys were combined to constitute one merged KFP survey: household survey, women’s survey and community survey. Questionnaires were designed for each survey separately and data collection was conducted in parallel in the same selected villages. The statistical model used for this study was bivariate analysis with chi-square tests to calculate the differences with a 95 % confidence interval, and multivariate logistic regression analysis. The tests were performed with STATA/IC 13.0.

11The data collection of the original surveys was conducted by National Institute of Statistics of Niger (INS) in collaboration with UNICEF. They were exempted from an ethical approval for human subjects given no implication of serological or biological test according to regulation no. 2004–011 on 30 March 2004.

12An inventory of mothers eligible for the communication program promoting KFP was established prior to the intervention. However, this database includes exclusively outcome indicators related to infant feeding and child healthcare and no data was collected regarding exposure variables such as socio-demographic and economic status of mothers. The information was not exploitable to compare before and after the intervention to measure the impact of the program activities on early initiation of breastfeeding. A post-intervention survey was therefore designed in a retrospective manner independently from this baseline dataset.

13Shortcomings of this cross-sectional survey should be noted. It is important to note that intervention sites were not randomly selected and subject to biases unlike a trial known to measure the impact of program interventions in the most accurate manner. From an epidemiological perspective, measurement bias with uncontrolled confounding effects over time is a major shortcoming of this cross-sectional study (Kopec et al., 1990). Causal relationships between an outcome variable and exposure variables could not be accurately determined due to the communication program being implemented over an extensive period of time.

Results

14The 2011 survey showed that 30 % of the mothers exposed to the program intervention were in polygynous unions (tab. 1). The proportion of polygynous unions was significantly higher in the exposed group than unexposed group (30 % vs. 23 %, p=0.002).

15As to regional disparity, Maradi, where the majority of the exposed group reside (n=325, 48 %), recorded the largest proportion of polygynous mothers among other 4 regions.

16Bivariate analysis showed that marital status was significantly associated with early initiation of breastfeeding (tab. 2): 66 % of mothers in polygynous unions initiated breastfeeding within the first hour of birth compared with those in monogamous unions (54 %, p=0.002). We examined what characterizes those married in polygynous unions in terms of socio-economic and demographic status of mothers and behavioral indicators that were identified as key determinants of early initiation of breastfeeding in the previously published article (Horii et al., 2017). They include: educational attainment and type of occupation of the interviewed mothers, decision of harvest use within a household, use of toilet facilities, support from neighbors or their parents to obtain transport to access the nearest health facility, access to the health facility. A part from decision of harvest use and access to the nearest health center, all variables showed statistically significant association with marital status (tab. 3).

17It appeared that the eldest age group over 21 years were the most likely to be in polygynous unions (29 % vs. 17 % in the youngest age group of 15-20 years, p<.001). Among mothers who gave birth more than once, those in monogamous unions reached 27 % vs. 16 % in primipara (p=.001). Secondary or higher school education was significantly associated with polygynous unions (32 %) whereas the proportion drops to 23 % among no education (p=.003). Household work with no direct income showed the lowest proportion of polygynous unions (22 %) compared to those in sales and services (44 %). Those who used improved latrine were more likely to be in polygynous unions compared to those who defecated openly (39 % vs. 24 %, p<.01). Among mothers who did more than 4 antenatal care (ANC) during the pregnancy, the proportion of polygynous unions reached 23 % whereas those who never did ANC included 32 % (p=0.004).

18We further examined these associations between the above-preselected determinants having shown statistically significant association and polygynous unions by taking into account confounding variables (tab. 4). A multivariate logistic regression analysis was conducted to measure the odds of being married in polygynous unions adjusted for age of mothers, parity, educational attainment, occupation, region and exposure to the program. Statistical significance disappeared for most variables associated with polygynous unions in the bivariate analysis except for few variables: the age group of 35-49 years are 1.8 times more likely to be in polygynous unions than those aged 15-20 (95 % CI: 1.1, 3.1). Sales and services as a principal activity was 1.9 times the odds of being married in polygynous unions as agriculture or livestock (AOR: 1.9, 95 % CI: 1.1-3.2). Use of toilets and number of ANC were no longer significantly associated with polygynous unions.

19Finally the association between early and exclusive initiation of breastfeeding and polygyny was measured (tab. 5). We measured a crude odds of early initiation of breastfeeding with regard to polygynous unions compared with these adjusted for all variables significantly associated with the outcome variable (early initiation of breastfeeding) (Horii et al., 2017) or the independent variable in question (polygynous union) earlier in the bivariate analysis (Tab. 3). In bivariate analysis, polygynous mothers are 1.7 times more likely to initiate breastfeeding within the first hour of birth than monogamous mothers (95 % CI: 1.3, 2.2). After having adjusted for socio-economic, demographic status of mothers and behavioral indicators, polygynous unions was no longer associated with early breastfeeding.

Table 1 – Distribution of Mothers in Polygynous Unions by exposure to the program

Table 2 – Early Initiation of Breastfeeding Practice within 1 hour after Delivery According to the Characteristics of Interviewed Mothers

Table 3 – Determinants of polygynous unions (chi2)

Table 3 – Determinants of polygynous unions (chi2)

Table 4 – Determinants of polygynous unions (adjusted odds)

Table 5 – Early initiation of breastfeeding according to marital status (crude and adjusted odds)

Discussion

20Distribution of polygynous unions is not equal and there exists significant difference between the exposed and unexposed groups. There are several reasons why the program rather extended its services to polygynous mothers: First, a majority of the exposed group of the survey is found in Maradi region where traditionally women are prone to polygynous marriage. This region has the largest proportion of polygynous unions (n=176, 38 %) among other regions (p<0.001). Another explanation is that as stated earlier we question the confounding effects of other factors. It is not because mothers are in polygynous unions but because they are better-off they readily accessed health facility and child care services made available by the program within the village. That is to say when they are richer, they get better chance to be provided support. The previous research revealed that socio-economically vulnerable mothers were more likely to be excluded from program support such as home visits, public debate led by community leaders and peer child healthcare promotion (Horii et al., 2016).

21The characteristics of the polygynous unions clearly indicated better-off socio-economic status. The results of bivariate and multivariate analysis of determinants of polygynous unions implied that multifaceted variables that determined marital status of mothers should be considered carefully when interpreting the likelihood of early initiation of breastfeeding by mothers in polygynous unions. Polygynous mothers tend to be over age 35, better-off because they could afford the improved latrine at home and benefit from stable income derived from sales and services they are most likely to be involved in. The assumption here is that these explanatory variables have strong influences on polygynous unions positively associated with early initiation of breastfeeding.

22Previously published findings of the statistical analysis suggested what determined early initiation of breastfeeding (Horii et al., 2017). Parity showed inconclusive findings regarding its association with postpartum breastfeeding (Waiswa et al., 2010; Narayan et al., 2005) and this reveals that even mothers who delivered for the first time could possibly put their child to the breast within the first hour of birth. As opposed to previous studies reporting no influence of education levels on optimal breastfeeding (Matanda et al., 2014), koranic school education was identified as a risk factor for delayed initiation of breastfeeding (Horii et al., 2017). However, the multivariate analysis showed that educational attainment did not match the characteristics of polygynous unions. Several variables of socio-economic status were associated with early initiation of breastfeeding and clearly showed early breastfeeding was impaired by income poverty as translated by household work with no direct income (Horii et al., 2017). Many polygynous are elder and doing sales and services. A few women we met in a village in Maradi said that they could spare time to do some business other than cooking and cleaning for the family by sharing household tasks between co-spouses. We may draw a conclusion that polygynous unions as it stands could not be a protective factor but it is the socio-economic status of mothers that determine the timing of initiating breastfeeding. The better-off households have more access to social and healthcare services and as they could afford it, they were keener to do peer healthcare promotion activities which showed the outstanding statistical significance on early initiation of breastfeeding among all other behavior change communication activities (Horii et al., 2016). Whereas when they struggle to survive day by day, it is very unlikely that they are directed to taking actively part of promoting healthcare as a volunteer to other mothers.

23The post-intervention survey did not investigate ethnic groups or religion of the interviewed mothers. According to the 2006 Niger Demographic Health Survey, the population is predominantly Muslim in the four regions (Maradi, Zinder, Tillabéry and Thaoua: n=1,975, 99 %). A study on Islamic ethnic groups in Guinea-Bissau by Gunnlangsson showed that colostrum feeding, considered to be culturally beneficial to newborn health, was widely practiced unlike other ethnic groups in Sub-Saharan Africa (Gunnlaugsson, Einarsdottir, 1993; Horii, 2015). Gunnlangsson’s study also described cultural abhorrence of multipara mothers’ first milk in the Islamic ethnic group. This cultural belief specific to Muslim opposes the assumption made by other studies that mothers having delivered many times in the past could more easily practice early and exclusive initiation of breastfeeding after birth. Withholding of colostrum is specific to cultural and traditional customs of each ethnic group and is subject to further investigations to explore the relation between ethnicity and socio-cultural factors related to postpartum breastfeeding in rural Niger. Besides the previously published article using the same dataset showed that whether being primipara or multipara did not change much the outcome of early breastfeeding (Horii et al., 2017). The determinant analysis in this chapter showed indeed that parity did not interfere with the association between polygynous mothers and early breastfeeding. This implies that behavior change communication could get over hindering factors related to cultural belief incompatible with optimal breastfeeding practices.

24Polygynous mothers who benefited from means of transport to get to the nearest health facility did not necessarily do more than 4 ante natal care (ANC). Being in polygynous unions, where mothers spared more time and were better off and better educated, was not itself a favorable environment to use preventive healthcare available at health facility during pregnancy. Previously published study showed the limited influence of health professional proving counseling during ANC at health center on promoting early breastfeeding (Horii et al., 2016). This implies that, although early breastfeeding is impaired by socio-economic vulnerability of mothers, income level does not alone lead mothers to practicing good behavior for healthcare during the postpartum period. Early initiation of breastfeeding could be determined by who are present and by what could be done at the time of delivery (Kirkwood et al., 2013). Unfortunately the survey includes no variable related to delivery assistance to examine the effects of actors assisting a majority of mothers delivering at home. A lesson-learnt could be drawn from an example of a participatory community based behavior change communication promoting Key Family Practices in Mali: traditional birth attendants, first reference point for pregnant women, attending home delivery in a village, played a key role to manage newborn care during the postpartum period. The program led by UNICEF was designed to involve these traditional birth attendants in referral system to let them accompany mothers to do ANC at a health center and to train them in neonatal and infant feeding and healthcare (Guitteye et al., 2010).

25Some literature shows conflicting outcomes about the relationships between co-spouses in polygynous unions: collaborative and supportive dialogue between women of reproductive age could contribute to behavior change of mothers for stopping suboptimal breastfeeding. On the other hand, polygynous households could often be prone to conflicts between co-spouses (Bove, Valeggia, 2009). The post-intervention survey does not provide any further information to explore how interactions between women of reproductive age married in polygynous unions can impact neonatal care and early breastfeeding after birth in rural Niger. The effect of polygynous unions on early initiation of breastfeeding appears to be inconsistent and we therefore conclude that polygyny is not a determinant of early initiation of breastfeeding independently of socio-economic factors.

26Economically deprived mothers are at greater risk of delaying initiation of breastfeeding. Association between polygyny and early initiation of breastfeeding is influenced by socio-economic vulnerability of mothers prone to suboptimal postpartum breastfeeding. A behavior change strategy should adopt socio-culturally acceptable methods and provides economically accessible and sustainable responses to address the poverty gap regardless of marital status of mothers, whether they are in monogamous or polygynous unions. Further research on typology of communication strategies should be conducted to investigate whether specific types of socio-economic inequality, ethnicity and religion sensitive actions would effectively address optimal breastfeeding promotion and lead a whole community to actively taking part in improving health care of their children.