Religious fasting often involves abstention from animal source foods (ASFs). Although children are exempt, their diets are influenced by the widespread fasting practices. This study investigated the factors influencing ASF consumption among young children during the Lent fasting period in western Amhara, Ethiopia. We used baseline survey data from households with children 6–23 months of age (n = 2,646). We conducted regression analysis to examine the maternal and household factors associated with ASF consumption and path analysis to examine the direct and indirect effects of maternal knowledge, beliefs, social norms, and livestock ownership on ASF consumption. Only 24% of children consumed any ASF in the previous day—18% dairy products, 5% eggs, and 2% flesh foods. Mothers with high knowledge, beliefs, and social norms about feeding children ASFs during fasting had higher odds (odds ratio: 1.3–1.4) of children who consumed them. Compared with households with no ASFs, those with ASFs available were 4.8 times more likely to have children who consumed them. Most of the association between knowledge, beliefs and social norms, and ASF consumption was explained by pathways operating through ASF availability (approximately 9, 12, and 8 pp higher availability, respectively), which in turn were associated with higher consumption. Cow ownership was directly and indirectly associated with ASF consumption, whereas having chickens was indirectly associated with consumption via the availability pathway. Our findings corroborate the importance of maternal behavioural determinants related to feeding ASFs to children during fasting on ASF consumption via household availability and the positive influence of livestock ownership.
Study data were drawn from the baseline household survey conducted as part of the programme evaluation of Alive & Thrive (A&T) Phase II (2015–2017) in Ethiopia. A&T is an initiative to save lives, prevent illness, and contribute to healthy growth and development through improving infant and young child feeding (IYCF) practices. In Phase I (2009–2014), A&T operated in Bangladesh, Ethiopia, and Vietnam, reaching millions of children less than 2 years of age through large‐scale social and behaviour change communication interventions and achieving considerable gains in IYCF practices (Kim et al., 2016; Menon, Nguyen, Saha, Khaled, Kennedy, et al., 2016a; Menon, Nguyen, Saha, Khaled, Sanghvi, et al., 2016b; Rawat et al., 2017). The focus of Phase II in Ethiopia is to operationalize the Government of Ethiopia’s National Nutrition Plan in one region, Amhara, to improve breastfeeding and complementary feeding practices using a multiple sector approach. The baseline survey was carried out in March and April 2015, during the Lent fasting period, in 20 woredas (districts) belonging to three western zones of Amhara region (Awi, North Gondor, and West Gojjam) that do not participate in the Productive Safety Net Programme (national cash and food transfer programme targeted to chronically food‐insecure households; Government of the Federal Democratic Republic of Ethiopia, 2004). Within each of the 20 survey woredas, four enumeration areas (EAs—a geographical unit consisting of 150–200 households) were randomly selected, to yield a total of 80 EAs. At each EA, households from a listing of all eligible households were randomly selected to meet the estimated sample sizes for the impact evaluation. A total of 2,646 households with children 6–23 months of age participated in the survey. Given that 96.1% of our sample were Orthodox Christian, we restricted our sample to this group only. Data on ASF consumption were available for 2,536 mother–child pairs. All mothers of the study children were provided with information about the study at recruitment, and verbal informed consent was obtained from all participants. Data were collected via face‐to‐face interviews using a structured questionnaire. Ethical approval was obtained from the Institutional Review Boards of Addis Continental Institute of Public Health in Ethiopia and the International Food Policy Research Institute in Washington, D.C., USA. Child feeding practices were assessed by asking mothers about all liquids and solid and semisolid foods consumed by their children during the previous day. Dietary data for children 6–23 months of age were categorized into seven food groups: grains, roots, and tubers; legumes and nuts; dairy products (milk, yogurt, and cheese); flesh foods (meat, fish, poultry, and liver/organ meats); eggs; vitamin‐A‐rich fruits and vegetables; and other fruits and vegetables. The main outcome variable was consumption of ASF (yes/no), which was defined as consumption of any of the following three food groups: dairy, flesh foods, and eggs in the past 24 hr. We also constructed five complementary feeding indicators on the basis of the WHO recommendations (WHO, 2010) including (a) minimum dietary diversity (received foods from 4 or more food groups), (b) minimum meal frequency as appropriate for age, (c) minimum acceptable diet (achieved the minimum dietary diversity and age‐appropriate minimum meal frequency), (d) consumption of iron‐rich foods, and (e) timely introduction of solid, semisolid, or soft foods (introduced to these foods when infants were 6–8 months of age). Because nearly all children were breastfed, we did not stratify the sample on breastfeeding status. We measured several potential behavioural determinants—maternal knowledge, beliefs, and perceived social norms (Glanz, Rimer, & Viswanath, 2008). Knowledge questions about ASF were drawn from a broader set of IYCF knowledge questions validated in a previous study (Menon, Ruel, Arimond, & Ferrus, 2003), whereas questions related to beliefs and social norms were developed specifically for this study on the basis of formative research findings (Alive & Thrive, 2016). Maternal knowledge about ASF was assessed on the basis of mothers’ responses to questions related to knowledge on timely introduction of different ASFs, frequency of feeding children ASFs, and benefits of ASFs to young children for growth and brain development (Table S1). Each correct answer was given a score of 1, yielding a total knowledge score of 10 (range 0–10). Maternal beliefs about ASF feeding during fasting was assessed on the basis of levels of agreement to statements about whether children should be fed eggs, milk, and meat during and outside the Lent fasting period. Each item was measured using a 5‐point Likert scale in which women responded with the degree to which they agreed or disagreed with the two statements (Table S1), yielding an overall score of 10 (range 2–10). Overall score (range 2–10) for perceived social norms about ASF feeding during fasting was created on the basis of the levels of agreement to two statements: (a) Most people who are important to me approve of me feeding eggs, milk, and meat to my child during Lent fasting and (b) most women who have young children like me feed their children eggs, milk, and meat during Lent fasting (Table S1). The scores for knowledge, beliefs, and social norms were used as continuous and categorical (using median cut‐off) independent variables. Because both methods showed similar results, we present the findings using the median as the cut‐off levels to compare those who have high scores with low scores in the behavioural determinants. Household availability of ASF was based on observation of whether there were any eggs, milk, or meat available in the house for feeding children by interviewers at the time of the survey. Mothers were also asked if the household owns any of the three types of livestock animals—chickens, cows, and goats or sheep. Selection of covariates were guided by previous literature on determinants of dietary patterns or diversity in low‐ and middle‐income countries (Mayen, Marques‐vidal, Paccaud, Bovet, & Stringhini, 2014) and in Ethiopia (Workicho et al., 2016). Variables that may be associated with child consumption of ASF were measured at child (age and sex), mother (age, education, and occupation), and household levels (food security and socioeconomic status [SES]). Household food security was measured using the Food and Nutrition Technical Assistance/United States Agency for International Development Household Food Insecurity Access Scale (Coates, Swindale, & Bilinsky, 2007). Household SES was created by principal components analysis using a set of items related to house and land ownership, housing quality (house construction materials), household assets (different types of durable goods and productive assets not including livestock previously mentioned), and access to utilities (water, electricity, gas, and sanitation services). The first component derived from component scores was used to divide household SES into tertiles (Vyas & Kumaranayake, 2006). Descriptive analyses were used to examine the characteristics of the study sample and complementary feeding practices including ASF consumption. Bivariate associations between predictor variables and ASF consumption were examined using chi‐square tests and unadjusted logistic regression models. Path analysis using the structural equation modelling command in Stata was conducted to examine the potential mediating effect of ASF availability on the relationship between maternal knowledge, beliefs, social norms, and livestock ownership on ASF consumption. Path analysis allows us to simultaneously estimate all regression equations identified in a model and quantify the direct and indirect effects of different independent variables on ASF consumption. The indirect effect on ASF consumption was calculated as the product of the two path coefficients: (a) between maternal behavioural factors/livestock ownership and ASF availability and (b) between ASF availability and ASF consumption. All models were adjusted for child, mother, and household covariates and geographical clustering at the EA level (Kline, 2011). The proportion of mothers reporting ASF consumption among children was within 0.2–0.8, so linear regression was used rather than logistic regression because results are essentially the same within this range (Cox & Snell, 1989; Hellevik, 2009). All the results were considered significant at P < 0.05. Data analysis was performed using Stata version 13.