This study has documented breastfeeding and infant feeding practices in two slum settings in sub-Saharan Africa. It has also identified the factors associated with sub-optimal breastfeeding and infant feeding practices in these two slum settings. The study finds that though there is almost universal breastfeeding, exclusive breastfeeding is rare. Complementary foods are initiated too early; only two percent of children were exclusively breastfed before the age of six months and the mean age of introduction of complementary foods was one month.
In our study, we see an overall picture of universal breastfeeding with the majority of infants breastfed for at least 12 months. These findings are similar to findings in other developing countries [9, 26]. However, this picture masks the unwelcome finding that the WHO recommendations with regards to breastfeeding and introduction of complementary food were largely not adhered to. Early initiation of breastfeeding following delivery, as recommended by the WHO is not universally being practiced in developing countries, despite the importance of colostrum in providing the baby with rich nutrients. The findings that close to 40% of the infants were not breastfed within one hour following delivery are in agreement with findings from studies in other developing countries including Uganda, India and Bangladesh [20, 26, 27]. In our setting, the main reasons stated for failure to introduce babies to breast milk immediately after birth were health-related including the mother having insufficient milk. This has also been documented in other studies . Other cultural factors have been noted in other settings such as Nigeria, for example the belief that colostrum is dirty milk, hence harmful to the baby, a belief that the mother should rest and clean up first, and performance of rituals and prayers before the baby starts breastfeeding . In our study though, cultural factors did not feature in the responses from the women, supposedly because our study was a quantitative, rather than qualitative study.
In this study complementary foods were initiated too early, despite two thirds of the women in our study being aware of the WHO recommended time to initiate complementary feeding. This is in line with previous studies conducted in rural Kenya, Malawi and Uganda [26, 29, 30], and in some other slum settings in the developing world [31, 32]. In a study conducted in the late 1990s on the determinants of child nutritional status in six African countries, Madise et al. reported very low levels of exclusive breastfeeding among infants under the age of 4 months with percentages ranging from two percent in Nigeria to about 34% in Tanzania . The few studies that have looked at slum settings particularly in Asia paint a similar picture [31, 32]. The finding showing the persistence of early introduction of complementary feeding in the region is critical given the importance of exclusive breastfeeding to child health. Exclusive breastfeeding protects against infections such as gastrointestinal and respiratory infections, and enhances motor development in the child [4, 34, 35]. In a study using data from Botswana which examined the association between breastfeeding, morbidity, and malnourishment, Chikusa (1991) found that children aged 4 months or younger who had been weaned had more than 11 times the odds of having diarrhea compared with those who were still being breastfed . The main reason cited for introducing complementary food early was the mother's perception of insufficient breast milk. This finding is in line with other studies from other settings which have shown that the perceived lack of sufficient breast milk is a main reason for early breastfeeding cessation or early introduction of complementary foods [31, 37, 38].
The variables associated with the cessation of breastfeeding during the first year of the child's life include the mother's marital status, her ethnicity, and her level of education, and perceived size at birth. The association between marital status and early cessation of breastfeeding has been reported in many studies with conflicting results [22, 23]. In this study, women who were not in union, particularly those who were formerly married were more likely to stop breastfeeding their infants than women who were in union. It has been suggested that the association between marital status and breastfeeding cessation may be due to the presence or absence of social, emotional and economic support of a partner ; however, these factors were not assessed in our study. A more plausible reason in Kenya, where HIV is high, is that a disproportionately large number of formerly married women are HIV positive and many women in this situation were until recently advised to exclusively breastfeed their infant for 6 months and then to rapidly wean .
The evidence of the association between a mother's level of education and the duration of breastfeeding also varies [20, 24, 25]. In this study, lower than secondary level education was associated with earlier cessation of breastfeeding. While it is not very clear why this is the case, higher education may be associated with higher knowledge and practice of positive health behaviour. Higher HIV prevalence among those with less than secondary level education, especially those with no education at all in our setting  may be associated with early cessation of breastfeeding. We also observed an association between ethnicity and breastfeeding cessation. All other ethnic groups apart from the Kamba were more likely to stop breastfeeding their infants compared to Kikuyu women. There is no established reason for this but it could be multi-factorial, including cultural practices related to breastfeeding and child rearing. Further, HIV prevalence, which may affect breastfeeding practices, has also been documented as higher amongst the Luo and Luhya ethnic groups in this slum setting compared to the Kikuyu ethnic group . Additionally, evidence from the study areas indicates that Kikuyus have lower fertility compared to other ethnic groups. Prolonged breastfeeding may explain or be explained by lower fertility. Mothers who get pregnant while breastfeeding are more likely to stop but, also mothers who breastfeed for longer period have lower chances of getting pregnant . Additionally, better child health outcomes have also been documented among the Kikuyu's compared to other ethnic groups in this study setting  and our findings may indicate that the Kikuyus have better health-related behaviours and practices than most of the other ethnic groups in the study area. The association between birth size and the duration of breastfeeding has not been studied in depth. Our study found that children who were perceived to be larger at birth were less likely to be stopped from breastfeeding earlier. This is similar to a US study, infants who were breast-fed for less than 4 months were smaller at birth than those who were breast-fed for 4 months or more . The factors behind this association in our study setting need further investigation.
Predictors of early introduction of complementary foods include the child's sex; the mother's marital status, her ethnicity, and her level of education; the desirability of the pregnancy of the index child, the place of delivery and the slum setting. Boys were more likely to be introduced to complementary feeding early compared with girls. Anecdotal evidence indicates that boys are introduced to complementary foods early because breast milk alone does not meet their feeding demands. Having never been in union/married was associated with higher risk of early introduction of complementary foods. A positive association between being married and exclusive breastfeeding has been documented in other studies . As in the case of the duration of breastfeeding, this may be associated with social, emotional and economic support of a partner . Similar to the finding related to the duration of breastfeeding, all other ethnic groups apart from the Kamba group, were more likely to initiate complementary foods earlier than the Kikuyus. This may be related to cultural practices and other factors such as HIV status as described for duration of breastfeeding. While a few studies have linked the mother's education with early introduction of complementary foods , similar to our study, the negative influence of a mother's low education on early introduction of complementary foods has been observed in many studies in other settings, suggesting a need for education and health promotion to change these harmful feeding practices [21, 45, 46].
Slum of residence, was associated with the timing of introduction of complementary foods. Mothers from Viwandani, were at lower risk of introducing complementary foods before six months. This may be because Viwandani, being in the industrial area and attracting labourers to the industries is likely to have more educated people (other than the mother) for example the father and other household members, who may affect infant feeding practices. Mothers who delivered at home were more likely to introduce complementary foods earlier than those who delivered in a health facility. Mothers who deliver in a health facility in most cases receive breastfeeding counselling, especially with the revitalisation of the Baby Friendly Hospital Initiative (BFHI) from 2007 aimed at promoting optimal breastfeeding practices. The BFHI has been found to be effective in several settings in the developing world . BFHI, is being revitalized in Kenya in the National Strategy on Infant and Young Child feeding , and it may be playing a role in encouraging mothers to exclusive breastfeeding their infants in the first 6 months of life. Since the BFHI initiative was introduced, there has been potential improvement in the proportion of children exclusively breastfed from 13% in 2003 to 32% in 2008 . The positive association between pregnancy desirability and complementary feeding has barely been previously studied. In our study, infants who were unwanted at conception but were wanted later were less likely to be introduced to complementary foods early. The association between pregnancy desirability and breastfeeding and complementary feeding practices needs further investigation.
Urban slum settings present unique challenges with regards to breastfeeding and infant and young child feeding practices due to their physical and socio-economic characteristics. In these informal settings, basic government services including health care services are limited and this, coupled with financial constraints, leads to a substantial proportion of women in these slums giving birth at home or at informal private health facilities [48, 49]. This means that most of these slum women are systematically excluded from government initiatives such as those aimed at promoting optimal breastfeeding and infant feeding practices, based at health facilities such as the BFHI mentioned above, which involves counselling of mothers on infant and young child feeding around the time of delivery. Another unique characteristic of slum settings is limited livelihood opportunities  hence food insecurity. As indicated in this study and in other slums such as in India , one key reason for initiating complementary foods too early is due to the mother having inadequate breast milk. While the important role of hormones and the psychosocial status of the mother in lactation is well established, though limited evidence exists, volume of milk produced may also be related to maternal nourishment. A review of breastmilk volumes and composition among poorly nourished communities indicated that milk volumes were lowest in communities with poor levels of nutrition and poor living conditions .
Potential interventions to address the unique challenges in the slum settings should address both access issues and the socio-economic limitations. A potential intervention to counteract the systematic exclusion from basic government services may include, home-based counselling of mothers on infant and young child feeding by community based health workers and/or supporting the (informal) private service providers for instance through training programs to offer services according to established government guidelines such as those on breastfeeding. The effectiveness of such interventions in health care delivery, including promotion of optimal infant feeding practices in resource-constrained settings has been indicated . To enhance adequacy of milk produced by the mothers, potential interventions may be to enhance maternal nourishment through ensuring food security. This may be through appropriate income generating activities to enhance livelihoods. Food supplementation has also been found to enhance breastmilk volume . Additionally, interventions that empower the new mother by demonstrating correct breastfeeding techniques, ways of stimulating breast milk production, and counselling on proper nutrition may improve breastfeeding practices .
Limitations in this study relate to missing values in some of the variables particularly the socio-economic status variables. Appropriate measures were taken in the analysis to minimize bias as indicated above. It would have been important to follow-up the children for a longer period to establish complete duration of breastfeeding in line with the WHO recommendation that breastfeeding should continue for two years or beyond. This was however not done and children were only followed up till they were slightly more than one year old. Despite these limitations, this study has key strengths that are worth mentioning. The study provides important information on infant breastfeeding and feeding practices in informal settings in sub-Saharan Africa, for which there is a dearth. A key strength of this study lies in its longitudinal nature, minimising recall bias that may be associated with cross-sectional studies. The study involved rigorous follow-up hence information for most of the children was updated by the end of the follow-up period, reducing bias due to loss to follow-up. The study involved a census of all children born to mothers in two defined geographical areas; hence there was minimal bias due to sampling error.