Background:  Understanding how early infant feeding practices influence subsequent 18-month HIV-free survival of HIV-exposed infants is important to inform clinical guidance; however, little information is available.

Methods:  HIV-infected pregnant women in KwaZulu Natal, South Africa, were enrolled into a non-randomized intervention cohort. Counselors supported mothers in their choice to either exclusively breastfeed for 6 months or to exclusively replacement feed from birth. Infant feeding data were obtained weekly from mothers; blood samples from infants were taken monthly at clinics to diagnose HIV infection. Kaplan-Meier analyses conditional on infant feeding practices were used to assess 18-month mortality, HIV transmission risk and HIV-free survival; associations with maternal and infant variables were quantified in Cox regression analyses.

Results:  We included 1193 live-born infants. Overall, 18-month probability of death (95%CI) was 0.04 (0.03 to 0.06) and 0.53 (0.46 to 0.60) for HIV-uninfected and -infected children, respectively; unadjusted hazard ratio 16.9 (11.5 to 24.8). The 18-month probability of survival among HIV-uninfected children (96%) was not statistically significantly different for infants breastfed (n = 800) or replacement fed (n = 128) from birth. Overall probability of HIV-free survival was 0.76 (0.73 to 0.78). In univariate analysis of infant feeding practices in the first 6 months of life, probability of HIV-free survival beyond that age (conditional on survival to 6 months) was 0.98 (0.89 to 1.00) in replacement fed from birth infants (reference), 0.96 (0.90 to 0.98) in infants breastfed for <6 months (p = 0.25) and 0.91 (0.87 to 0.94) in children breastfed for >6 months (p = 0.03). Breastfeeding beyond 6 months increased the risk of HIV acquisition:  adjusted hazard ratio (AHR) 3.3, 1.0 to 10.5, p = 0.05 compared to breastfeeding for <6 months. In multivariable analysis allowing for infant’s HIV infection, with replacement-fed from birth as reference, the probability of survival beyond 6 months of age did not statistically significantly differ by breastfeeding duration:  infants breastfed for <6 months AHR 1.2, 0.1 to 11.9, p = 0.86; breastfeeding for >6 months AHR 0.9, 0.1 to 7.1, p = 0.92.

Conclusions:  Breastfeeding of HIV-uninfected infants beyond 6 months of age increases the risk of HIV acquisition without gains for survival. It remains important, however, to identify means of making breastfeeding safer for HIV-infected women who have no choice than to continue breastfeeding.