Racial and Ethnic Differences in Breastfeeding

Abstract

Objectives: Breastfeeding rates differ among racial/ethnic groups in the United States. Our aim was to test whether racial/ethnic disparities in demographic characteristics, hospital use of infant formula, and family history of breastfeeding mediated racial/ethnic gaps in breastfeeding outcomes.

Methods: We analyzed data from the Community and Child Health Network study (N = 1636). Breastfeeding initiation, postnatal intent to breastfeed, and breastfeeding duration were assessed postpartum. Hierarchical linear modeling was used to estimate relative odds of breastfeeding initiation, postnatal intent, and duration among racial/ethnic groups and to test the candidate mediators of maternal age, income, household composition, employment, marital status, postpartum depression, preterm birth, smoking, belief that "breast is best," family history of breastfeeding, in-hospital formula introduction, and WIC participation.

Results: Spanish-speaking Hispanic mothers were most likely to initiate (91%), intend (92%), and maintain (mean duration, 17.1 weeks) breastfeeding, followed by English-speaking Hispanic mothers (initiation 90%, intent 88%; mean duration, 10.4 weeks) and white mothers (initiation 78%, intent 77%; mean duration, 16.5 weeks); black mothers were least likely to initiate (61%), intend (57%), and maintain breastfeeding (mean duration, 6.4 weeks). Demographic variables fully mediated disparities between black and white mothers in intent and initiation, whereas demographic characteristics and in-hospital formula feeding fully mediated breastfeeding duration. Family breastfeeding history and demographic characteristics helped explain the higher breastfeeding rates of Hispanic mothers relative to white and black mothers.

Conclusions: Hospitals and policy makers should limit in-hospital formula feeding and consider family history of breastfeeding and demographic characteristics to reduce racial/ethnic breastfeeding disparities.

FIGURE 1

Mediation model pathways. Path C and C′ differ in that path C is the raw association between the independent and outcome variables, and path C′ is the residual association between the independent and outcome variables after the indirect effect of the mediator(s) is included in the model. Only racial/ethnic contrasts significant at Path C were pursued. Of those, mediation analysis was discontinued if Path A was not significant.

FIGURE 2

Summary of possible mediation relationships. Working, postpartum depression, and smoking were only tested with duration because they were measured at 1 month postpartum, which occurs after breastfeeding initiation and postnatal intent were measured in the hospital. BF, breastfed; WIC, Special Supplemental Nutrition Program for Women, Infants, and Children.