A missing step: Assessing breastfeeding initiation rates at the community level

Background. Many breastfeeding initiatives begin at the state or federal level, success requires sustained local implementation. Understanding the local factors influencing breastfeeding rates warrants community wide data rather than institutional (e.g. WIC) or birth certificate data. Local data more clearly identify gaps in programs and outcomes. Methods. Analyses of population level breastfeeding initiation we employed a modified Pregnancy risk Assessment Monitoring surveillance methodology. We assessed differences between low (LIM; Medicaid funded delivery or prenatal WIC enrollment) and non-low income (NLIM) mothers. Over 1000 mothers (50.9% LIM) completed the survey (births May 2009-May 2010) representing 6,139 mothers after weighting. Three logistic regression models (all participants, LIM-only and NLIM-only) identified independent predictors of not initiating breastfeeding. Potential confounders (e.g. education, age, stresses, prenatal care start, intendedness, insurance, abuse, birth outcomes) were identified in bivariate analyses (p≤0.1) and included in a backwards selection model building process. Results. Initiation was relatively high (82.3%). In the first model (all participants) income was not significant but unmarried, Black race, multiparous, fewer pre-pregnancy preventive practices and less overall prenatal provider counseling were risk factors for non-initiation. Additionally in the LIM stratified model Hispanic and more pre-pregnancy positive health behaviors increased initiation. By contrast only three factors in the NLIM model were significant: financial stress increased initiation while multiparous and Hispanic decreased initiation. Conclusions. Using local, granular data we identified risk factors for not initiating breastfeeding particularly among low income mothers . These findings have implications for pre/interconception care given that few are modifiable during pregnancy.