ABSTRACT
The Breastfeeding, Antiretrovirals, and Nutrition (BAN) Study randomized HIV-infected mothers and their infants to receive either maternal lipid-based nutrient supplements (LNS) during lactation or no LNS and then to 1 of 3 antiretroviral drug (ARV) arms (maternal, infant, or no drugs). Assigned interventions were provided from 0 to 28 wk and all infants (n = 1619) were given LNS during (24-28 wk) and following (28-48 wk) weaning. This paper assesses the feasibility of infant LNS as a breastmilk replacement and uses longitudinal random effects models to examine associations of interventions, morbidity, and season with weight-for-age (WAZ), length-for-age (LAZ), and BMI-for-age (BMIZ) Z-scores from 24 to 48 wk. Infant LNS adherence was high (94.1% ate it daily). From 24 to 48 wk, mean WAZ (-0.42 to -0.76 SD; P < 0.001) and LAZ (-0.93 to -1.56 SD; P < 0.001) steadily declined, whereas BMIZ remained >0 throughout. A higher LAZ was associated with assignment to the maternal LNS arm (ß=0.19; P < 0.05). Lower WAZ and BMIZ were associated with seasonal food insecurity (ß=-0.08 and -0.09, respectively; both P < 0.001), fever (ß=-0.07 and -0.13; both P < 0.001), diarrhea (ß=-0.19 and -0.23; both P < 0.001), and assignment to the infant ARV arm (ß=-0.17 and -0.17; both P < 0.05). The magnitude of the season and morbidity effects was small and BAN infants had higher weights and lengths than their counterparts in the general population. High LNS adherence and the modest impact of morbidity on growth indicate that LNS is a feasible breastmilk replacement for HIV-exposed infants weaned early, but controlled trials are needed to quantify the effects of LNS on growth in this population.
