Menu
Menu
Menu
Menu
Dietary diversity in children may be influenced not only by individual circumstances but also by the features of the community in which they live. Our study aimed to assess community and individual-level determinants of minimum dietary diversity among children aged 6-23 months in Ethiopia. We included 2960 children aged 6-23 months from the recent Ethiopia Demographic and Health Survey. A minimum dietary diversity was defined as the consumption of at least five food groups out of the eight reference food groups within 24 h by children aged 6-23 months. Multilevel logistic regression was used to investigate the drivers of minimum dietary diversity in Ethiopian children aged 6-23 months. About 125 % of children met the bare minimum of dietary diversification. Age of the child (9-11 months AOR, 33 (95 % CI 18, 56), 12-17 months AOR, 40 (95 % CI 24, 67), 18-23 months AOR, 35 (95 % CI 20, 58)), caregiver listening radio at least once a week AOR, 16 (95 % CI 11, 24) and wealth quantiles (Second AOR, 18 (95 % CI 11, 31), Fourth AOR, 29 (95 % CI 16, 52) and Highest AOR, 22 (95 % CI 11, 42)) were individual characteristics associated with dietary diversity. Place of residence was the only community-level characteristic associated with children’s dietary diversity (Rural AOR, 04 (95 % CI 02, 06)). The minimum dietary diversity among Ethiopian children is suboptimal. Nutrition programmes aimed at enhancing dietary diversity should be strengthened in this population, particularly for those from poor families and residing in rural areas.
The study was carried out in Ethiopia, a country located in Northeastern Africa. The country has a total estimated population of 109⋅2 million people and covers about 1⋅1 million square kilometres of area and has great geographical diversity, ranging from 4550 m above sea level to 110 m below sea level. The data were collected based on the country’s previous nine administrative regions and two administrative cities(17), but the country now has two additional regions (Sidama region and South West Ethiopia Region) that are separate from the Southern Nations, Nationalities and Peoples’ Region (SNNPR). The administrative region is divided into zones, districts, towns and kebeles (the smallest administrative units). The present study used a cross-sectional, secondary data analysis design. We used the most recent and nationally representative 2016 Ethiopian Demographic Health Survey (EDHS) data(17). A stratified two-stage cluster sampling technique was applied. A total of 645 enumeration areas (EAs) were chosen in the first stage, using probabilities proportionate to EA size (202 in urban and 443 in rural) (based on the 2007 EPHC frame). A fixed number of twenty-eight households in each cluster were chosen using an equal probability systematic sampling technique in the second stage. For this study, Kids Record (KR) file containing information about women and children was used, and important variables related to inadequate dietary diversity were extracted from the dataset. In the present study, 2960 weighted data of children aged 6–23 months were used for analysis. Based on the updated WHO guideline(8), minimum dietary diversity was defined as the proportion of children aged 6–23 months who consumed at least five food groups out of the eight referenced food groups within 24 h. These food groups are (1) breast milk; (2) grains, roots, and tubers; (3) legumes and nuts; (4) dairy products; (5) flesh foods (meats/fish/poultry); (6) eggs; (7) vitamin A-rich fruits and vegetables; and (8) other fruits and vegetables. The total dietary diversity score ranges from 0 to 8, with 1 point given for each of the 8 food groups consumed. Children with dietary diversity scores ≥5 were classified as they attained the minimum dietary diversity, whereas those with scores <5 were classified as unmet MDD. The outcome variable was coded as 1 for adequate dietary diversity and 0 for inadequate dietary diversity. We selected possible determinants based on evidence from literature and the availability of variables in the EDHS-2016. We investigated the effect of explanatory factors on dietary diversity at both the individual and community levels. The study included individual-level determinants such as child, maternal and paternal characteristics. The children's characteristics included sex, age (in months), birth order and episodes of cough or fever in the last 2 weeks. Maternal characteristics included: age (years), highest educational level, frequency of listening to the radio, frequency of watching television, attending Antenatal care (ANC) follow-up, place of delivery, postnatal care visit and maternal empowerment(22). Paternal characteristics included paternal characteristics including the highest educational level and occupation. Household characteristics include the household wealth index, the gender of the household head, the number of children under the age of five and the number of total household members. Parents’ occupations were classified as Not working (unemployed), Nonagricultural works (professional, technical, managerial, clerical, sales, services, skilled manual and unskilled manual), Agricultural works (agricultural – employee) and others. The community-level determinants included contextual region (agrarian dominant, city dwellers dominant and pastoralist dominant), place of residence (either urban v. rural) and aggregate variables such as community poverty (higher v. lower), community distance to a health facility (distance a big problem v. distance, not a big problem) and remoteness of the location. Community poverties were created from mean values of wealth index categories of the individual mothers for each cluster. The two values for the community poverty level were higher poverty and lower poverty. The EDHS did not capture data that can directly describe the characteristics of the community/clusters except the place of residence, mean rainfall, mean temperature and altitude. Hence, we created community variables by aggregating the individual-level characteristics within their clusters. The aggregates were computed using the average values of the proportions of women in each category of a given variable. Likewise, based on the national median values aggregate values were categorised into groups. These aggregate community-level determinants include contextual region, community distance to a health facility and remoteness of the location. We used GIS estimates of travel time to cities to construct a ‘living in a remote location’ dummy variable that equals 1 if the DHS cluster has more than a one-hour travel time to a town/city of 20 000 people or more(22). Contextual region: For this study, the administrative regions were categorised into agrarian, pastoralist and city, based on their settings that may have a relationship with child dietary diversity. Since regions used for administrative purposes might not necessarily be related to the child feeding practice of the population. The regions of Tigray, Amhara, Oromia, SNNPR, Gambella and Benshangul-Gumuz were recorded as agrarian. The Somali and Afar regions were combined to form the pastoralist region and the city administrations – Addis Ababa, Dire Dawa and Harar – were combined as a city. Though Gambela and Benshangul-Gumuz have been considered pastoralists in recent times, their living settings approached the agrarian(23). Ecological level variables such as mean rainfall per year, 1985–2015 (mm), mean temperature, 1985–2015 (Celsius) and altitude (metres) were also included. The data were analysed using STATA version 16 (StataCorp, College Station, TX, USA). In this analysis, households with children aged 6–23 months old with no missing information on dietary diversity were included. To adjust for the redistribution of samples to different regions and the possible variation in response rates, we used sampling weight in all the analyses. The ‘Svy’ command was used to allow for adjustments for the cluster sampling design. Categorical variables were reported using absolute and relative frequencies; whereas continuous variables were summarised using mean with standard deviation (sd) or median and interquartile range (IQR) for variables that deviate from normal distribution after visual examination using a histogram. Due to the nature of the EDHS data, being a hierarchical structure, data are often correlated and thus cannot be assumed, independent. Hence, to identify individual and community-level determinants of dietary diversity, we performed a multilevel logistic regression. A multilevel approach adequately adjusts the unexplained variability of the nested structure and can estimate cluster-level effects on the outcome variable. Therefore, in the present study, a two-level mixed-effect logistic regression analysis was employed to estimate the independent (fixed) effects of the explanatory variables on dietary diversity adjusting for cluster and regional-level random effects. To investigate the community and individual-level determinants of minimum dietary diversity among children aged 6–23 months, any variable with a P-value of 0⋅25 on a univariable test was a candidate for the multivariable model, along with all variables of known clinical importance. Four models were fitted and compared. Model 1 was an empty model which was fitted without independent variables to test random variability using the Intraclass correlation coefficient (ICC); Model 2 include individual-level factors (age of the child, mother's educational level, frequency of listening a radio and wealth quintile of household); Model 3 include community-level factors (place of residence (either urban v. rural), mean annual rainfall of the cluster 1985–2015 (mm) and mean temperature 1985–2015 (Celsius)) and Model 4 include both individual and community-level factors. The relative fits of these models were then compared using the Akaike (AIC) and Bayesian Information Criterion (BIC), and the difference in model fit was compared using the χ2 test. Finally, the adjusted odds ratios (AORs) with the 95 % confidence intervals (95 % CI) were reported. We requested access to the datasets from the Demographic and Health Surveys (DHS) program/ICF International, and permission was granted by DHS program data archivists to download the dataset for this study. Before the authors could access the data, it was de-identified. The data were only used for the registered research topic and were not shared with anyone else.
N/A
