This study aimed to document whether food insecurity was associated with beliefs and attitudes towards exclusive breastfeeding (EBF) among urban Kenyan women. We conducted structured interviews with 75 human immunodeficiency virus (HIV)-affected and 75 HIV-status unknown, low-income women who were either pregnant or with a child ≤24months and residing in Nakuru, Kenya to generate categorical and open-ended responses on knowledge, attitudes and beliefs towards EBF and food insecurity. We facilitated six focus group discussions (FGD) with HIV-affected and HIV-status unknown mothers (n=50 women) to assess barriers and facilitators to EBF. Of 148 women with complete interview data, 77% were moderately or severely food insecure (FIS). Women in FIS households had significantly greater odds of believing that breast milk would be insufficient for 6months [odds ratio (OR), 2.6; 95% confidence interval (95% CI), 1.0, 6.8], that women who EBF for 6months would experience health or social problems (OR, 2.7; 95% CI, 1.0, 7.3), that women need adequate food to support EBF for 6months (OR, 2.6; 95% CI, 1.0, 6.7) and that they themselves would be unable to follow a counsellor’s advice to EBF for 6months (OR, 3.2; 95% CI, 1.3, 8.3). Qualitative analysis of interview and FGD transcripts indicated that the maternal experience of hunger contributes to perceived milk insufficiency, anxiety about infant hunger and a perception that access to adequate food is necessary for successful breastfeeding. The lived experience of food insecurity among a sample of low-income, commonly FIS, urban Kenyan women reduces their capacity to implement at least one key recommended infant feeding practices, that of EBF for 6months. © 2010 Blackwell Publishing Ltd.
The study was conducted between August and November 2008, in Nakuru, a multi‐ethnic, urban municipality, where EBF is commonly promoted in antenatal clinics (ANC), maternity wards, health centres and through media. Nakuru is located located approximately 160 km northwest of Nairobi in the Great Rift Valley Province and is the fourth largest city (population approximately 300 000) in the Republic of Kenya. Rates of EBF are low in Kenya – 18% and 13% at 4 and 6 months, respectively, in spite of ongoing promotion of EBF to 6 months in health centres and through the media and moderate rates of prolonged breastfeeding (57% to 20–24 months) (UNICEF 2008). In 2007, the HIV prevalence rate among adults aged 15–49 years was 7.4%, with women being disproportionately infected (8.7%) compared with men (5.6%) (NASCOP 2009). The prevalence of food insecurity is also relatively high. In 2005, the Kenya National Bureau of Statistics (KNBS) estimated that 51% of Kenyans were undernourished (KNBS 2008) 1 . This proportion has likely increased following recent events including crop failures due to drought, the post‐election violence of 2008 and the global food crisis. In Nakuru, the HIV prevalence in 2009 was 5.9% [females (6.6%); males (3.5%)] (http://www.aidskenya.org/public_site/webroot/cache/article/file/Official_KAIS_Report_20091.pdf), and a recent survey reported that approximately 75% of HIV‐affected households in Nakuru were moderately to severely food insecure (Mbugua 2009). This study used cross‐sectional quantitative and qualitative data to investigate associations between indicators of food insecurity and attitudes and beliefs about EBF. Because the study was interested in breastfeeding attitudes and beliefs of urban women who are or would soon be breastfeeding, participants were sampled based on recent or near‐future breastfeeding experience. Thus, sampling for interviews and focus group discussions (FGDs) was purposive based on the following criteria: either confirmed pregnant by a health care worker or with a child ≤24 months of age, resident in Nakuru municipality and expecting to reside in the municipality for at least the following year. Women were sampled equally from two groups according to HIV status as described below. We recruited 75 women from ‘HIV‐affected’ households who met the eligibility criteria to participate in structured interviews. We defined women from ‘HIV‐affected’ households as those whose households could be identified at baseline to have at least one adult member diagnosed with HIV through membership of a community‐based organization for ‘living positively’ with assistance from a well‐respected, community‐based key informant (gatekeeper). We recruited the first 30 of these women by convenience from among the participants in a longitudinal follow‐up assessment conducted by the authors on the effects of an urban agriculture intervention on infant and young child nutrition among HIV‐affected households in Nakuru (the Sustainable Environments and Health Through Urban Agriculture, or SEHTUA project) (Karanja et al. 2010). We recruited the remaining 45 of these women from community‐based HIV/acquired immunodeficiency syndrome (AIDS) support/self‐help groups using service‐based convenience sampling. We recruited an additional convenience sample of 25 women meeting the eligibility requirements from Prevention of Mother to Child Transmission Services (PMTCT) at the government‐run Provincial General Hospital (PGH) in Nakuru to participate in three FGDs. Women in this group were not required to disclose to researchers their own HIV status or that of any member of their household during data collection. We recruited 100 low‐income women attending ANC or child well clinics (CWC) at the PGH (75 for interviews and 25 for FGDs). A maternal and child health nurse approached women in the ANC or CWC waiting area to inform them of the study and to invite participation. Those agreeing were referred to a trained researcher for eligibility screening, consent and administration of questionnaires. It was assumed for analysis that most women in this group lived in less HIV‐affected households, i.e. most households did not contain any adult members diagnosed with HIV/AIDS. Women in this group were not required to disclose to researchers their own HIV status or that of any member of their household during data collection. We used ‘mixed methods’ to collect a combination of qualitative and quantitative data through (1) FGD and (2) interviews with structured and open‐ended responses. All interviews and focus groups were conducted in Kiswahili using guides developed in English and pre‐tested in Kiswahili with mothers in a town approximately 1 h away from Nakuru. All FGDs and approximately 25% of structured interviews were recorded, transcribed and translated into English and reviewed by the co‐authors. A research assistant with a university degree in nutrition was trained on the interview tool and FGD guide prior to and during piloting of both instruments; training included reviewing the questions, rapport building and interview skills such as probing. During training, a bilingual co‐investigator with extensive training in infant and young child feeding and qualitative methods provided feedback and retraining. During the implementation of the study, rapid review of transcripts allowed for retraining as needed. We used FGDs facilitated by a trained researcher in a private setting at the PGH to gather community‐based barriers to EBF. We followed an FGD guide that asked open questions about barriers and supports to EBF in the community and perceptions on different sources of infant feeding information. Specific questions on food insecurity were not asked. However, if themes related to food insecurity were independently raised by women, the facilitator probed on this topic. All FGDs were recorded and later transcribed into English. We conducted interviews at a location of the participant’s choosing (Group A: usually the home but also in private spaces at community‐based comprehensive care centres or support groups) or in a quiet and private location at the PGH (Group B). The interview guide was semi‐structured and designed to collect individual knowledge, attitudes and practices towards EBF using both closed‐ended questions to elicit pre‐coded categorical responses and open‐ended questions that allowed for follow‐up and probing. In the first part of the interview, structured questions were designed to capture quantitative socio‐demographic, obstetric and food insecurity indicators as follows: In the second part of the interview, we used a modified set of open‐ended responses previously used in another town in Kenya (Sellen 2006) to explore personal experiences of breastfeeding in the context of food insecurity and hunger; responses were further probed for detail and meaning using follow up questions and probes. We also used a modified set of semi‐structured questions on infant feeding knowledge, attitudes and beliefs, and experiences with infant feeding counselling based on previous work conducted by team members in Voi, Kenya (Sellen et al. 2007). Study protocols and instruments were approved by institutional review boards at the University of Toronto and the Kenyan National Council of Science and Technology and the Medical Director of the PGH in Nakuru. All participants provided informed consent prior to participating. We categorized participants’ households into two food insecurity categories: ‘food secure/mildly food insecure’ or ‘moderately/severely food insecure’, using HFIAS responses, coding guidelines provided by the Food and Nutrition Technical Assistance Project (Coates et al. 2007) and SPSS 15.0 (SPPS‐IBM, Chicago, IL, USA). We tested statistical models predicting an association between food security status and beliefs and attitudes towards EBF using multivariate logistic regression specifying household food security status in the preceding month as a predictor and categorical indicators of current attitudes and beliefs towards EBF as outcomes. We adjusted models for the following variables as determined by changes in the odds ratio (OR) (Kleinbaum & Klein 2002): maternal age, parity (nulliparous/multiparous), years of schooling, pregnancy status (yes/no), respondent contributes to household income (yes/no); household size; whether the mother received infant feeding counselling (yes/no). Significant effects were determined at P < 0.05. We also tested for interactions between food security status and receipt of infant counselling for attitudes and beliefs towards EBF (P < 0.10 cut‐off for significance). Analyses were conducted using SAS v9.1 (SAS Institute, Cary, NC, USA). We conducted subgroup analysis stratified on HIV group status to assess the extent of potential confounding. This revealed that there was insufficient variance in the HIV‐affected group to examine associations because 97% of respondents were moderately to severely food insecure. Subgroup analysis of HIV‐status unknown participants only yielded similar trends as when the groups were combined, although estimates did not reach statistical significance due to smaller sample sizes (available on request). Thus, we report findings for the entire sample of HIV‐affected and HIV‐status unknown women with group status (HIV affected or status unknown) removed from the model due to multi‐collinearity with food insecurity status. Household income was also not included in the adjusted models due to multi‐collinearity and because it was missing for a substantial portion of participants (30%) due to inability to estimate or uncertainty regarding monthly household income. We conducted content analysis of open‐ended responses from the interview (attitudes, beliefs towards EBF) and of FGD transcripts using post hoc thematic categorization and manual coding and labelling of open‐ended response. Common themes around beliefs and attitudes towards EBF were identified and reviewed by the authors.
N/A