While literature describing immigrant’s breastfeeding practices exists, especially among those living within developed countries, there is a significant gap in knowledge on how the host culture may influence the EBF behaviors of refugees, especially those living in protracted situations within sub-Saharan Africa. A cross-sectional study was conducted in the Buduburam Refugee Settlement in Ghana from July-August 2008 to explore the association between the amount of time living in Ghana and exclusive breastfeeding practices among Liberian refugees and Ghanaians in surround villages. The study included 480 women: 239 Liberians living in 12 settlement zones (in two of which Liberians and Ghanaians co-exist), 121 Ghanaians living in two settlement zones, and 120 Ghanaians living in nearby urban village of Awutu. Liberian mothers who lived in Ghana at least eight years were significantly more likely to exclusively breastfeed (OR: 1.78, 95% CI: 1.02, 3.09) compared to Ghanaian mothers living in Awutu. These findings suggest that increased time living in Buduburam improved the chances of EBF success among Liberians, perhaps as a result of unique EBF education/support opportunities offered in the settlement to Liberian refugees that were not readily available to Ghanaians. Further research to understand the “mechanisms” explaining exclusive breastfeeding differences as a function of time spent in host country is needed for improving breastfeeding support in refugee settlements and host communities.
The data presented are drawn from the quantitative survey administered in a larger mixed methods study designed to identify shifts in diet, food access and availability, and cultural beliefs among Liberian refugees living in Buduburam Refugee Settlement. The larger mixed methods study focused on three objectives: (a) understanding changes in dietary practices of Liberians living in the settlement; (b) assessing food availability, food access, dietary practices, and cultural beliefs of Liberians and the relationship with nutritional status; and (c) assessing how Liberians influenced these dietary factors among Ghanaians living in Buduburam or in a surrounding village. To achieve the main study objectives: (a) an in‐depth interview was administered to two to three generations of Liberian refugee and Ghanaian caretakers living within and around Buduburam and (b) a quantitative survey was administered to Liberian refugee and Ghanaian women living within and around Buduburam. The mixed methods study setting included Buduburam and the adjacent urban village of Awutu, located 5 km from the settlement. Buduburam is composed of 12 zones, with zones 1–10 being settled predominately by Liberian refugees. Zones 11–12 include the original village of Buduburam and the outlying areas where both Liberians and Ghanaians households coexist. Infant feeding practices were only examined through the quantitative component of the study; thus, findings from the qualitative component are not presented in this article. Liberians and Ghanaian female caretakers living in Buduburam and Awutu village were recruited to participate in the quantitative survey component of the study. Women were eligible for the study if they met the following criteria: (a) Liberian or Ghanaian, (b) at least 16 years old, (c) having a biological child that was between 6 and 59 months at the time of the survey, (d) not being pregnant, (e) not having health problems or conditions that would cause any changes in their diet, and (e) must have lived in either Buduburam or in an Awutu village located less than 5 km from Buduburam. If there were multiple children in the household between 6 and 59 months, the youngest child was the index child. A cross‐sectional survey was administered between July and August 2008. The survey included demographic, socioeconomic, household food security, infant feeding practices, maternal dietary intake, acculturation, and maternal health status questions. The survey was first pilot‐tested with five Liberians and four Ghanaian women of the target population, and revised accordingly. The survey was administered to 480 women from all 12 Buduburam settlement zones and Awutu following a systematic sampling approach (i.e., selection of every fifth house). Within Buduburam, 12 interviews were conducted in each zone except for zones two (13 interviews) and eight (11 interviews). Within each zone, the central location was first identified. Standing within that central location, interviewers randomly selected a direction and visited the first house observed. Interviewers then moved to every fifth house to recruit and interview participants. If the house was empty, interviewers revisited the house until they received a reply. If an occupant of the house did not meet eligibility, interviewers continued onto the next fifth house. Interviewers continued interviewing within each zone until the required sample size was achieved. For interviews conducted in Awutu, a local villager assisted interviewers in locating houses and introducing residents. The same sampling method was used of selecting every fifth house. The final sample included 120 Liberians from zones 1–10, 119 Liberians from zones 11–12, 121 Ghanaians from zones 11–12, and 120 Ghanaians from urban, Awutu villages 5 km from Buduburam. Four interview teams, each composed of one Ghanaian and one Liberian, conducted all interviews. The Liberian interviews were administered by Liberians in Liberian pigeon English and Ghanaian interviews were administered by Ghanaians in either English or a local Ghanaian language. Interviewers, who had previous experience with anthropometry and nutrition surveys, were from the local communities and received an intensive 3‐day training in survey administration and standardized anthropometric measurement prior to beginning data collection. Interviewers were closely supervised by study investigators, and the data collected were reviewed daily to maintain quality control. Prior to survey administration, verbal consent was obtained from all participants. Participants were assured that all information obtained would be confidential, would be used only for research purposes, and that their participation would not affect their access to nutrition or food assistance programs in the camp, nor would be used for determining repatriation. The study received Institutional Review Board approval from the University of Connecticut and the University of Ghana. Additional approval to conduct data analysis came from Yale University. Permission to defer ethical approval to the collaborating Universities was provided by representatives of Buduburam Refugee Settlement. Data were de‐identified and maintained in a secure, locked file cabinet by the principal investigator. The main outcome variable was whether or not mothers’ EBF the index child for the first 6 months of life. Mothers were asked if they breastfed the index child, and when water or other liquids were first introduced to the index child. Similarly, mothers were asked when solids or semisolid foods were first introduced to the index child. Responses for both multiple choice questions ranged from less than 1 month to greater than 7 months. If the mother indicated that either water and/or liquids/solids/semi solid foods were introduced before 6 months, then that child was classified as not EBF (coded as 0). If the mother indicated that both water and liquids/solid/semi solid foods were introduced at 6 months or later, then that child was classified as EBF (coded as 1). The main four‐level independent acculturation proxy variable was computed from data on the mother’s: nationality, time spent living in Ghana, and where she lived in Ghana (for Ghanaians). Nationality was defined by the mother’s self‐report of her nationality, Liberian or Ghanaian. Time spent living in Ghana (excluding any time abroad) was asked of all participants. If mothers spent time abroad, they were asked to identify which countries and length of time abroad. To assess acculturation, two categories were created for Liberians using the median length of time living in Ghana (8 years) among Liberians: Liberians living in Ghana less than 8 years and Liberians living in Ghana at least 8 years. Location of residence was not used to assess acculturation among Liberian refugees because of the following: (a) EBF rates were very similar between Liberians living in zones 1–10 and 11–12 (70% and 66%, respectively) suggesting location did not influence EBF among this population, and (b) the variable identifying the zone where Liberians lived was not considered a sufficient proxy for acculturation, which typically includes variables such as language preferences, place of birth, nativity, and length of time in the new country (Pérez‐Escamilla & Putnik, 2007). Ghanaians were classified based on where they lived. Ghanaians living in zones 11–12 were classified as such, while Ghanaians living in Awutu comprised another comparison group to identify if any differences in EBF behavior existed between those living within or outside of the camp. The resulting four‐level independent acculturation proxy variable included the following: (a) Liberians living in Ghana less than 8 years, (b) Liberians living in Ghana at least 8 years, (c) Ghanaians living in zones 11–12, and (d) Ghanaians living in Awutu. To validate this variable as a proxy for acculturation, it was crossed with other acculturation proxies (i.e., birth country, country raised in, and environment raised in) and was found to be highly correlated. Key covariates that were assessed included: maternal age, age of index child, maternal body mass index (BMI), parity, and social capital. Maternal age was self‐reported. The exact age of the index child was obtained by using the date that the survey was administered and the child’s date of birth obtained from his/her weighing and vaccination card. Maternal BMI was obtained from measured maternal weight and height. Parity was assessed using the question, How many children have you given birth to? Social capital was obtained by asking, During the past 12 months, have you borrowed money from any neighbors, family members, or friends? Maternal and child anthropometric measurements were taken using weight, height, mid‐upper arm circumference, and head circumference (index child only) following standard recommended procedures. Maternal health status was assessed by asking mothers to rate their own health in the last 6 months. Food insecurity over the past 6 months was assessed using an adapted version of the 16‐item Latin American & Caribbean Household Food Security Scale (Pérez‐Escamilla et al., 2009). This scale was pretested in Buduburam and since then has been validated globally. It is the source of the Food and Agriculture Organization Food Insecurity Experience Scale, which has been psychometrically validated in over 150 countries (FAO, 2016). All statistical analyses were conducted with SAS 9.3 (SAS Institute Inc., 100 SAS Campus Drive, Cary, NC, USA). Chi‐squared tests were used for bivariate analyses of categorical variables, and t‐tests were used for analyses of continuous variables. Differences were considered significant at p < 0.05. Unadjusted odds ratios and 95% confidence intervals were used to report the bivariate associations between key covariates and EBF. Variables with a p‐value <0.10 in bivariate analyses were included in the full multivariate binomial logistic regression model to identify significant predictors of EBF. Stepwise backward elimination was used to specify the final model. This involved manually removing a non‐significant variable at each step and assessing changes in the model likelihood ratio. After removing a nonsignificant variable, if a decrease in the likelihood ratio was observed and this difference was less than the chi‐squared critical value, then the variable was eliminated with each step. Factors known to be associated with EBF including age of the index child, parity, mother's BMI, and mother's age were kept in the final model (Pérez‐Escamilla et al., 1995). The final model included mother's BMI, mother's age (years), child age (months), parity, borrowed money from neighbor/family in last year, and the four‐level acculturation proxy variable combining nationality and time living in Ghana. Interactions between pairs of independent variables in the multivariate model were tested, but none were found.
N/A