Minimum acceptable diet and associated factors among infants and young children aged 6-23 months in Amhara region, Central Ethiopia: Community-based cross-sectional study

Objective The main objective of this study was to assess the prevalence of a minimum acceptable diet (MAD) and associated factors. Design Community-based cross-sectional study Setting Debre Berhan Town, Ethiopia. Participants An aggregate of 531 infants and young children mother/caregiver pairs participated in this study. A one-stage cluster sampling method was used to select study participants and clusters were selected using a lottery method. Descriptive statistics were calculated for all study variables. Statistical analysis was performed on data to determine which variables are associated with MAD and the results of the adjusted OR with 95% CI. P value of <0.05 considered statistically significant. Primary outcome Prevalence of MAD and associated factors Results The overall prevalence of MAD was 31.6% (95% CI: 27.7 to 35.2). Having mother attending secondary (adjusted OR, AOR=4.9, 95% CI: 1.3 to 18.9) and college education (AOR=6.4, 95% CI: 1.5 to 26.6), paternal primary education (AOR=1.3, 95% CI: 1.5 to 2.4), grouped in the aged group of 12-17 months (AOR=1.8, 95% CI: (1.0 to 3.4) and 18-23 months (AOR=2.2, 95% CI: 1.2 to 3.9), having four antenatal care (ANC) visits (AOR=2.0, 95% CI: 1.0 to 3.9), utilising growth monitoring (AOR=1.8, 95% CI: 1.1 to 2.9), no history of illness 2 weeks before the survey (AOR=2.9, 95% CI: 1.5 to 6.0) and living in the household with home garden (AOR=2.5, 95% CI: 1.5 to 4.3) were positively associated with increase the odds of MAD. Conclusion Generally, the result of this study showed that the prevalence of minimum acceptable was very low. Parent educational status, ANC visits, infant and young child feeding advice, child growth monitoring practice, age of a child, a child has no history of illness 2 weeks before the survey, and home gardening practice were the predictors of MAD. Therefore, comprehensive intervention strategies suitable to the local context are required to improve the provision of MAD. A community-based cross-sectional study design was used from February to March 2018 in Debre Berhan Town, North Shewa, Central Ethiopia. The Town is located 130 km away from Addis Ababa, the capital of Ethiopia. In 2017, the Town has an 88 369 total population, nine kebeles, one referral hospital, three health centres and 14 health posts. All infants and young children 6–23 months with mother/caregiver living in Debre Berhan Town and randomly selected kebeles were considered as the source and study population for this study, respectively. A household considered eligible if the infant and young child aged 6–23 months with mother/caregiver living for at least 6 months in the selected kebeles were included in the study. The child was excluded from the study if the mother/caregiver was absent in the household or the mother/caregiver was unable to respond because of the child’s illness or her illness and if the eligible household was closed after three revisits. The required sample size for the study was calculated using a single population proportion formula with the following assumptions: the proportion of MAD among infants and young children was 7%,17 a margin of error of 3%, power of 80%, 95% confidence level, design effect of 1.5% and 10% non-response rate. The final required sample size for the study was 459 infants and young children aged 6–23 months with a mother/caregiver. A one-stage cluster sampling method was used to select the study population. The town consists of nine kebeles (the smallest administrative unit in Ethiopia) and kebeles were considered as a cluster to select the study population. Among the nine clusters, three clusters were randomly selected and data were collected from every unit in the selected clusters. The total number of eligible infants and young children aged 6–23 months with their mother/caregiver was taken from the health extension workers (HEWs) record. According to the HEWs record, a total of 577 infants and young children aged 6–23 months—mother pairs were lived in the selected kebeles/clusters. Even though the final calculated sample size required for the study was 459 infants and young children aged 6–23 months pair with mother/caregiver, the total number of 577 infants and young children with mother/caregiver lived in the selected kebeles/clusters. Because of the nature of the cluster sampling method, all eligible (577) infants and young children aged 6–23 months with their mother/caregiver living in the selected clusters were included in the survey. Sociodemographic data were collected using pretested and structured interviewer-administered questionnaires developed from prior studies.11 20–22 The study was conducted per the Declaration of Helsinki ethical principles for medical research involving human subjects and each study participant gave informed written consent. The mother/caregiver was interviewed and used as a primary source of data for the study, but if the mother was absent, caregivers were interviewed to collect the data for the study. Ten diploma and two BSc nurses were trained for data collection and supervision, respectively. Data regarding the household wealth were collected using information from ownerships available assets; ownership of livestock, agricultural land, electronics, radio, television, refrigerator, car, bicycle, cart, gold, sofa, source of water, availability of electricity, type of toilet and household characteristics; type of wall, floor, and ceiling.17 Household food security was measured using Household Food Insecurity Access Scale (HFIAS), a validated tool developed by Food and Nutrition Technical Assistance. The HFIAS is based on respondent recall in the past 30 days and asks two closely related questions; nine occurrence questions that examine the experience of food insecurity in the past 4 weeks with two response choices as 1=yes or 0=no. Each occurrence questions followed by a frequency of occurrence question that asks the respondent how often the specific condition occurs in the past 4 weeks with the form of Likert scale response as 1=rarely (1–2 times in the past 30 days), 2=sometimes (3–10 times in the past 30 days) and 3=often (>10 times in past 30 days). When summing up the frequency of occurrence questions, the HFIAS score of household range 0–27 and severity of household food insecurity increase with increase the HFIAS score.23 IYCF practices were collected using WHO IYCF standardised questionnaires based on the mother recall of food groups given to her child 24 hours before data collection.10 Finally, all foods that the child consumed were grouped into seven food groups: (1) grains, roots and tubers; (2) legumes and nuts; (3) dairy products; (4) flesh foods; (5) eggs; (6) vitamin A-rich fruits and vegetables and (7) other fruits and vegetables.10 A household that did not experience any food insecurity conditions or just experience worry, but rarely in the past 4 weeks.23 A household that experiences one of the three levels of food insecurity conditions; mildly, moderately and severely food insecurity or access conditions in the past 4 weeks categorised as food insecure.23 Consumption of four or more food groups from the WHO recommended seven food groups within 24 hours day or night before the survey.10 The minimum number of times the child consumes solid, semisolid or soft foods (including two milk feeds for non-breastfed children) within 24 hours day or night before the survey. The minimum number of times is two times for breastfed children aged 6–8 months, three times for children aged 9–23 months, and four times for non-breastfeed children 6–23 months of age.10 Consumption of the MDD and MMF within 24 hours day or night before the survey.10 Providing a child with solid, semisolid or soft foods in addition to breast milk at the age of 6 months.10 A proxy measure of living standards derived from information on ownership available assets and household characteristics and household classified into terciles category.17 The explanatory variables used for determinant analysis were selected based on similar studies11 15 24 and the following variables were selected to identify factors associated with MAD. Age of mother categorised as: 19–24, 25–29 and ≥30 years of age; educational status of mother: no formal education, primary education, secondary education and college and above; occupational status: housewife, employed, merchant and farmer; mother involvement in deciding on what a child to be feed: involved or not involved; mother has a history of illness within 2 weeks before the survey: yes or no; antenatal care (ANC) visits during pregnancy: less than three ANC visits and four and above ANC visits; maternal fruit and vegetable consumption per week: consume less than three times per week and consume four or more times per week; mother received IYCF advice HEWs: yes or no; mother use child growth monitoring and promotion: yes or no; mother with a history of illness 2 weeks before the survey; and place of delivery: home delivery or health facility delivery. Father educational status: have no formal education, primary, secondary and college or above and father occupation: employed, merchant and farmer. Child sex: male or female; child age: age 6–11 months, age 12–17 months and age 18–23 months; child initiated to complementary feeding: yes or no; child age at which child introduced with complimentary food: <6 months, at 6 months and after 6 months; child currently bottle feed: yes or no and child has a history of illness with 2 weeks before the survey: yes or no. A household wealth index was constructed based on principal component analysis and the household was categorised into terciles: poor, medium and rich; head of household or a person who is responsible for decision-making in a household: father, mother or both; household food security; food secure and food insecure; the presence of home garden: yes or no; and family size: categorised ≤3, 4–5 and ≥6 family members. MAD was categorised into a dichotomous variable: meeting MAD=1 and not meeting MAD=0. A child who meets both the MDD and MMF was classified as meeting MAD otherwise classified as not meeting MAD. Data collection tools were initially prepared in English and translated into Amharic and then back to English to check for its consistency. A pretest was done on 5% of the study sample, 2 days of training were given for data collectors and supervisors. The principal investigator and supervisors have supervised the data collection process. Data were double-entered for cross-validation. First, data were checked for accuracy and completeness. Then, data were entered into Epi-Data V.3.1 and exported to SPSS V.22 for analysis. A Strengthening the Reporting of Observational Studies in Epidemiology cross-sectional reporting checklist was used.25 Descriptive statistics were used to describe sociodemographic, child feeding practice and maternal and child healthcare unitisation variables. Frequency and percentage were calculated for categorical data and the mean with SD was calculated for continuous variables. Multicollinearity between explanatory variables was checked with SE; a variable with a SE of ≥2 was dropped from the analysis. To select the appropriate analysis method between cluster-level analysis and ordinary logistic regression for a cluster sampling method, first, we fitted a null model and examined community variation or random effects. The measure of community variation (random-effects) was estimated with intra-class correlation coefficient (ICC) and the ICC result was 3%. Since the community variation was less than 5%, the use of an ordinary logistic regression analysis model is sufficient instead of a cluster-level analysis. Bivariable logistic regression analysis was done to assess the association between each covariate with MAD. Covariates with p value<0.25 during bivariable logistic regression analysis; parent education, maternal fruit consumption, head of household, IYCF advice from HEWs, ANC follow-up, growth monitoring utilisation, age of a child, a child has a history of illness 2 weeks before the survey, presence of home garden, household food security and wealth index were included in a multivariable logistic regression model to control all possible confounders and to identify factors significantly associated with MAD. Unadjusted and adjusted ORs with a 95% CI were calculated to estimate the strength association of each explanatory variable with MAD and if the percentage difference between unadjusted and adjusted OR of a variable greater than 10%, a variable considered confounder. Variables with p value<0.05 in the final model were declared statistically significant. A two-factor product term was used to test interaction effects and p value of<0.05 was considered significant.