Racial/Ethnic Variation in Breastfeeding Across the US: A Multilevel Analysis from the National Survey of Children’s Health, 2007
- First Online:
- Cite this article as:
- Belanoff, C.M., McManus, B.M., Carle, A.C. et al. Matern Child Health J (2012) 16(Suppl 1): 14. doi:10.1007/s10995-012-0991-1
- 679 Downloads
We examined whether differences across states in race/ethnicity-specific breastfeeding rates are due solely to state differences in individual factors associated with breastfeeding or additionally, certain state “contextual” factors. Using data from the 2007 National Survey of Children’s Health, multilevel models examined whether state variability in race/ethnicity specific breastfeeding initiation and duration to 6 months were explained by (1) individual sociodemographic characteristics of women in states, and (2) an aggregate state measure of the availability of evidence-based maternity care services related to breastfeeding. Observed variability of race/ethnicity-specific breastfeeding rates was only minimally reduced after adjusting for sociodemographic characteristics (Median Odds Ratios (MOR), breastfeeding initiation: non-Hispanic White = 1.46, non-Hispanic Black = 2.26; Hispanic = 1.89. MOR, breastfeeding for 6 months: non-Hispanic White = 1.36, non-Hispanic Black = 1.84; Hispanic = 1.56). Overall variability in the degree of state gaps changed little in adjusted models (breastfeeding initiation: non-Hispanic Black σ2 = 0.74, se 0.28, Hispanic σ2 = 0.45, se 0.11; breastfeeding to 6-months: non-Hispanic Black σ2 = 0.41, se 0.10, Hispanic σ2 = 0.22, se 0.05). The measure of maternity care services was positively associated with breastfeeding overall but generally did not explain a substantial portion of between-state variability nor the overall variability in racial/ethnic gaps. Contextual sources of variation in state breastfeeding practices and disparities remain poorly understood. Differences in the socioeconomic makeup of states do not fully explain variability. The association of state breastfeeding rates and disparities with relevant policy and practice factors should be further investigated.