Menu
Menu
Menu
Menu
Vitamin A deficiency is common among preschoolers in low-income settings and a serious public health concern due to its association to increased morbidity and mortality. The limited consumption of vitamin A-rich food is contributing to the problem. Many factors may influence children’s diet, including residential food environment, household wealth, and maternal education. However, very few studies in low-income settings have examined the relationship of these factors to children’s diet together. This study aimed to assess the importance of residential food availability of three plant-based groups of vitamin A-rich foods, household wealth, and maternal education for preschoolers’ consumption of plant-based vitamin A-rich foods in Addis Ababa. A multistage sampling procedure was used to enroll 5467 households with under-five children and 233 residential food environments with 2568 vendors. Data were analyzed using a multilevel binary logistic regression model. Overall, 36% (95% CI: 34.26, 36.95) of the study children reportedly consumed at least one plant-based vitamin A-rich food group in the 24-h dietary recall period. The odds of consuming any plant-based vitamin A-rich food were significantly higher among children whose mothers had a higher education level (AOR: 2.55; 95% CI: 2.01, 3.25), those living in the highest wealth quintile households (AOR: 2.37; 95% CI: 1.92, 2.93), and in residentials where vitamin A-rich fruits were available (AOR: 1.20; 95% CI: 1.02, 1.41). Further research in residential food environment is necessary to understand the purchasing habits, affordability, and desirability of plant-based vitamin A-rich foods to widen strategic options to improve its consumption among preschoolers in low-income and low-education communities.
Addis Ababa is the largest city in Ethiopia, and one of the fastest-growing cities in the African continent [29], with an estimated population of 3.6 million, 10.6% of which are estimated to be children under five years of age [30]. The population highly heterogeneous with regard to economic status [31], with an unemployment rate of 31.4%, and 18.9% of the population living below poverty line. The city is also the largest urban recipient of migrants [32,33]. Together, the services (63%) and industry (36%) sectors share almost all economic structure of the city [33]. The food environment in Addis Ababa is diverse, with vendors ranging from micro-vendors (locally known as Gulit) to formal supermarkets [34]. At the time of data collection, Addis Ababa was administratively divided into 10 sub-cities and 117 woredas (the smallest formal administrative unit), with each sub-city comprising of 10–15 woredas. This study utilizes data collected in the EAT Addis survey which collected data from households and vendors in residential areas in all woredas of Addis Ababa. Data collection was conducted in two rounds to account for seasonality, capturing July to August 2017, reflecting a wet season, and January to February 2018, reflecting the post-harvest period. We used a multistage sampling procedure. Each of the 117 woredas in Addis Ababa were included in the survey. Each woreda was divided into five clusters to simulate the enumeration procedure in the Ethiopian Demographic and Health Surveys [30]. One cluster from each woreda was selected for inclusion using a simple random sampling procedure. To identify eligible households with children under five years of age in the cluster, we used a systematic random sampling procedure visiting every third household from a random starting position until we reached a total of 60 households per cluster. These households were assessed for eligibility based on presence of at least one child under five years of age. In cases where an eligible household had more than one eligible child, one child was randomly selected to be the index child from whom dietary data was to be collected. The mother or caretaker of the index child was invited to participate in the survey and was also the main respondent at the household data collection. Mothers or caregivers not at home following three recruitment visits were declared unavailable. For the purpose of this study, household with children age below six months were not included. To identify a residential food environment representing the cluster, one household per cluster was randomly selected to serve as an index household, around which all vendors within a five-minute walk in all direction were surveyed. In total, 14 type of vendors such as kiosk, micro vendor, bakery, fruit/vegetable shop, four mill, butcher, cooperative shop, ET-Fruit, street food vendor, mini market, dairy shop, mobile micro-vendor, livestock market, and fish market were included in the survey. Vendors which were not open at the time of the visit were declared unavailable. Instruments were developed to collect data from households and vendors. These instruments were drafted in English and subsequently translated into Amharic. All data collectors and supervisors had extensive previous field research experience and received training on the data collection tools, interview procedures, and ethical conduct pertaining to the study [35]. Pilot testing refined the development of our survey tools and procedures, including the use of tablets with Open Data Kit (ODK) software. Primary data was sent directly to a protected data server at Addis Continental Institute of Public Health. An overall aim in the EAT Addis survey was to evaluate the association between food availability in residential food environments and food consumption of the household and pre-school children. To facilitate comparison, we developed a common metric to assess food availability and food consumption. A frequently used type of indicator is diet diversity, and indicators have been developed for use in different population groups including children [36,37,38]. We developed a set of food groups which could be used to derive diet diversity indicators both for the household and the preschoolers, as well as to be used to define availability in the residential area. For this particular study we used collected information on three plant-based vitamin A-rich food groups: vitamin A-rich fruits, dark green leafy vegetables, and vitamin A-rich vegetables and roots. The household survey featured data collection on a number of aspects, but of relevance for this study is mainly infant and young child feeding, household wealth, maternal education, and distance between households and index household. Children aged 6–59 months were included in this study. The caregiver of the index child was asked to recall the child’s food consumption during 24 h recall period. In addition to the initial recall by the caretaker a photo gallery of locally available items was used to augment the listing of food groups to ensure a common understanding among our enumerators and participants. For the purpose of this study, three photos with common items representative of the food groups were used: vitamin A-rich vegetables (pumpkin, carrot, red bell pepper), vitamin A-rich fruits (mango, papaya), and dark green leafy vegetables (amaranth, cassava leaves, Ethiopian kale, cabbage, Swiss chard, broccoli). Consumption was categorized as yes or no to each of the three food groups. The household wealth index was developed by principal component analysis and categorized into wealth quintiles [39]. This was calculated based on household assets, household characteristics, access to utilities, and infrastructure variables. Maternal education was assessed based on the reported highest level of grade completed by mothers at the time of the survey, and then categorized according to the Ethiopian educational system—never attend school/not finished first grade, grade 1–4, grade 5–8, grade 9–12, and college-educated [40]. Food security was assessed by use of household food insecurity access scale [41]. Distance between households and index household were calculated based on geographical positioning system (GPS coordinates) of each household during the data collection. The Euclidian distance was calculated based on the geographical location points (latitude/longitude) using a trigonometric approach (distance formula) [42]. A survey of all food vendors within a five-minute walking radius from the index household was conducted. The data collection tool for the residential food environment survey consists of the vendor characteristics, vendor properties, and food availability. The survey tool was pilot tested in three randomly selected residential food environments to assess the feasibility and the relevance of the terminology and typologies. The same food groups and photo gallery of locally available food items used in the household survey was also used in the survey of vendors in the residential area. Of relevance for this study, each vendor was categorized as selling or not selling any items from each of the three vitamin A food rich food groups, such as vitamin A-rich fruits, dark green leafy vegetables, and vitamin A-rich vegetables and roots. For each food group, the residential availability was defined as the presence of at least one vendor in the area selling any item from that particular group [43]. We used STATA 14 software for data analysis. Frequencies and percentages were calculated to report descriptive statistics. The predictor of plant-based vitamin A-rich food consumption was assessed using a multilevel binary logistic regression analysis with meaningful nested hierarchy at household and residential food environment level [44]. The intra-cluster correlation coefficient (ICC) in an empty model showed variability in child consumption of plant-based vitamin A-rich food attributed to differences in residential food environment, also referred as between-cluster variability. Initially, each independent variable was evaluated individually to generate unadjusted effect estimates. After this, three multivariable multilevel logistic regression models were fitted. Model I included residential food environment variables (level 2 variables), comprised of availability of vitamin A-rich vegetables and roots, dark green leafy vegetables, and vitamin A-rich fruits. Model II included household and individual variables (level 1 variables), comprised of household and individual status. Thus, Model I adjusted for residential wealth in tertial and household distance to the indexed household, and Model II adjusted for maternal age, child age, marital status, and number of under-five children [45]. Model III included both level 2 and level 1 variables together to adjust for maternal age, child age, marital status, number of under-five children, household distance to the indexed household, and residential wealth tertial. The observed associations were expressed as unadjusted and adjusted odds ratio with 95% confidence intervals. This study was approved by the Ethical Review Board of Addis Continental Institute of Public Health (Ref No. ACIPH/IRB/004/2015), and the Ethical Review Board at University of Gondar (R.No.-V/P/RCS/05/355/2019). No identifying information was available for this study.
N/A
