Socio-demographic and community-level determinants of delayed breastfeeding initiation in Borama, Somalia: a multilevel mixed-effects analysis

Background Timely initiation of breastfeeding is a proven life-saving intervention for reducing neonatal morbidity and mortality. Despite global advocacy, adherence in fragile health systems remains critically low. This study assesses the magnitude and multi-level epidemiological predictors of delayed breastfeeding initiation among postnatal mothers in Borama, Somalia. Methods An institution-based cross-sectional study was conducted among 392 postnatal mothers (aged 15–49 years) recruited via systematic random sampling at the Sheekh Osman Center for Community Health. Data were gathered using a pre-tested, structured interviewer-administered questionnaire. Delayed breastfeeding initiation was defined as initiating suckling more than 30 min postpartum. Participants were nested within 14 recognized municipal neighborhood clusters in Borama. To account for community-level clustering, a four-step hierarchical multilevel mixed-effects logistic regression was performed. Fixed effects were expressed as Adjusted Odds Ratios (aOR) with 95% Confidence Intervals (CI), and random effects were quantified using the Intraclass Correlation Coefficient (ICC). Results The magnitude of delayed breastfeeding initiation was 40.8% (95% CI: 35.9–45.8%; n = 160 ). In bivariate analysis, delayed initiation was significantly associated with maternal education level ( p < 0.001 ) and absence of antenatal breastfeeding information ( p < 0.001 ). In the final fully adjusted multilevel model, the absence of breastfeeding counseling during antenatal care was the strongest independent predictor of delay (aOR = 7.93, 95% CI: 4.51–13.95, p < 0.001 ). Mothers without formal education had three-fold higher odds of delayed initiation compared to college-educated mothers (aOR = 3.07, 95% CI: 1.33–7.07, p = 0.008 ). Mothers aged 15–20 years had lower odds of delay compared to those older than 30 years (aOR = 0.39, 95% CI: 0.18–0.87, p = 0.021 ). The null model demonstrated significant community clustering (ICC = 18.1%, p < 0.001 ), which increased to 25.6% in the final adjusted model. Conclusion Over four in ten mothers experienced delayed breastfeeding initiation. Gaps in antenatal counseling, lack of maternal formal education, and neighborhood-level clustering are primary drivers of delayed initiation. Health authorities must implement facility-based Baby-Friendly Hospital Initiative protocols and intensify antenatal lactation counseling.