The Role of Breastfeeding in Racial and Ethnic Disparities in Sudden Unexpected Infant Death

A Population-Based Study of 13 Million Infants in the United States

Melissa Bartick; Alexis Woods Barr; Lori Feldman-Winter; Mònica Guxens; Henning Tiemeier


Am J Epidemiol. 2022;191(7):1190-1201. 

In This Article


In our sample, the overall SUID rate was 0.91 (n = 11,942) per 1,000 live births. The SUID rate was 0.82 (n = 6,150) for NHW infants, 1.84 (n = 3,870) for NHB infants, 2.32 (n = 267) for AI/AN infants, 0.22 (n = 149) for Asian infants, and 0.55 (n = 1,506) for Hispanic infants. The overall breastfeeding rate at hospital discharge was 81.9%. For NHW infants, the breastfeeding rate was 83.0%; for NHB infants, 69.5%; for AI/AN infants, 74.7%; for Asian infants, 90.1%; and for Hispanic infants, 86.8%.

Table 1 shows the study population characteristics by breastfeeding status. The characteristics indicative of social disadvantage were disproportionately associated with not-breastfeeding for most covariates. Web Table 1 shows the study population characteristics by breastfeeding status within each racial/ethnic group. NHB and AI/AN mothers had a particularly high rate of unmarried status among not-breastfeeding mothers. Antenatal smoking was highest among not-breastfeeding AI/AN and NHW mothers (28% and 26%, respectively). Asian mothers who breastfed their infants were nearly equally likely to be US born as Asian mothers who did not breastfeed (approximately 16% each). This was not the case among Hispanic mothers (60% of those who did not breastfeed were US born vs. 47% who breastfed were US born), nor among NHW and NHB mothers.

Table 2 demonstrates the association between race/ethnicity and SUID. Compared with NHW infants, the largest group, AI/AN and NHB infants both had the highest adjusted odds ratios (aORs) of SUID: 1.42 (95% CI: 1.25, 1.60) for AI/AN infants; 1.40 (95% CI: 1.34, 1.46) for NHB infants; Hispanic infants had the lowest aOR: 0.71 (95% CI: 0.66, 0.75).

Table 3 lists the unadjusted and adjusted risks of not-breastfeeding for SUID overall and stratified for each racial/ethnic group. The risk for not-breastfeeding on SUID overall, adjusted only for race/ethnicity, was 2.22 (95% CI: 2.13, 2.30). The overall aOR for not-breastfeeding on SUID was 1.14 (95% CI: 1.10, 1.19). The most influential variables contributing to the adjusted effect estimate were maternal nativity, smoking, marital status, and education, in that order, when analyzed using individual variable elimination regressions. The aOR between not-breastfeeding and SUID was highest for Hispanic infants (1.29; 95% CI: 1.14, 1.46) and AI/AN infants (1.21; 95% CI: 0.93, 1.57), lower in NHW infants (1.16; 95% CI:1.09, 1.22), and lowest in NHB infants (1.07; 95% CI: 1.00, 1.14) and Asian infants (1.00; 95% CI: 0.62, 1.60). The AdjRD between not-breastfeeding on SUID was similar in all groups except for AI/AN infants, for whom AdjRD was much higher, but the CI was large, given its small absolute number (Table 3), and there was no significant risk difference among Asian infants.

Table 4 shows the association of race/ethnicity (relative to NHW) with SUID that was mediated by not-breastfeeding. In NHB infants, 2.3% of the racial disparities of SUID were mediated by not-breastfeeding, in which a lower breastfeeding rate relative to NWH infants partially explained their higher SUID risk. In Hispanic infants, 2.1% of the racial disparities of SUID was mediated by not-breastfeeding. However, this mediation was in the opposite direction: the higher breastfeeding rates of Hispanic infants relative to NHW infants partially explained their lower SUID risk. There was a small negative mediation effect for AI/AN and Asian infants. Web Table 2 demonstrates that AI/AN mothers breastfeed more than NHW mothers, if we account for background characteristics. Web Table 2 also shows that the higher likelihood of Asian mothers to breastfeed relative to NHW mothers becomes inverted once background factors are accounted for (the table shows the risk of not-breastfeeding). Figure 2 graphically illustrates Web Table 2, using triangles so the reader can visualize the complex interplay of risks.

Figure 2.

Triangles using separate regression models for race/ethnicity on sudden unexpected infant death (SUID), mediated by not-breastfeeding, United States 2015–2018. Non-Hispanic White infants were the referent. A) Triangle for non-Hispanic Black infants (NHB). B) Triangle for non-Hispanic American Indian/Alaska Native (AI/AN) infants. C) Triangle for Asian infants. D) Triangle for Hispanic infants. aOR, adjusted odds ratio; OR, odds ratio.

A sensitivity analysis of infants who died when younger than 4 months of age (i.e., 49.9% of all cases) showed similar results to those reported in Table 3 and Table 4 (Web Table 3 and Web Table 4), with the exception of Hispanic infants. In Hispanic infants, the aOR of not-breastfeeding on SUID was lower (1.17) than for the total sample (1.29) and no longer statistically significant (95% CI: 0.98, 1.41).