Individual and community level determinants of minimum meal frequency among breastfeeding children aged 6-23 months in Ethiopia: a multilevel analysis of 2019 Ethiopian Demographic Health Survey data

Minimum meal frequency (MMF) is a vital indicator of whether a child's energy needs are being met. Previous studies in Ethiopia on MMF have primarily focused on individual factors, often using basic logistic regression models with limited scope and small sample sizes. In contrast, this study aims to identify the key determinants of MMF among breastfed children aged 6-23 months in Ethiopia, examining both individual and community-level factors through multilevel logistic regression analysis, utilizing nationally representative data. The study utilized a community-based cross-sectional design with a sample of 1,245 6-23 month breastfed children in Ethiopia, using data from the 2019 Ethiopian Mini Demographic Health Survey. Mixed effect logistic regression was used to identify factors associated with MMF. Adjusted odds ratios (AOR) with 95% confidence intervals were used to assess variable associations, while Intra-cluster correlation (ICC), median odds ratio (MOR), and proportional change in variance (PCV) were employed to gauge random variation. The proportion of 6-23-month breastfeeding children with MMF was found to be 55.9% (95% CI: 53.2, 58.6). In mixed effect logistic regression; breastfeeding initiation time within days after birth [AOR = 0.44; 95%CI: (0.24, 0.80)], maternal primary educational level [AOR = 1.46; 95%CI: (1.03, 2.05)], maternal higher level education [AOR = 2.43; 95%CI: (1.22, 4.81)] and children from wealthiest household wealth index [AOR = 2.13; 95%CI: (1.04, 4.33)] were individual-level factors and children from city-based regions [AOR = 1.75; 95%CI: (1.10, 2.77)] was community level factors significantly associated with MMF. The final model indicates that 50.8% of the variation in MMF was explained by combined factors at both individual and community levels and also the variation across clusters remained statistically significant (ICC = 8.1% and variance = 0.29 with