I applaud Rollins et al. for the careful work that went into their study of HIV and infant feeding recently published in AIDS. Msellati and Van de Perre's  editorial emphasizes the conclusion that very high-quality counseling and follow-up result in similar levels of HIV-free survival in children in Africa whether breastfeeding is avoided from birth or continues until 6 months, at least if it is exclusive. Other studies in the region have come to similar conclusions, but Rollins et al. followed up breastfeeding longer than most. In addition, they achieved higher rates of exclusive breastfeeding than most. We are told that infants who were exclusively breast-fed ‘in the first 6 months tended to do better’ but no data on this are provided. (As virtually all exclusive breastfeeding takes place ‘in the first 6 months,’ perhaps they meant ‘for’?)
Although Rollins et al. imply it, Msellati and Van de Perre do not emphasize the fact that only when high-quality healthcare is available (such as Rollins et al. provided, as must all longitudinal studies, for ethical reasons), are we likely to see the pattern seen here whereby feeding pattern continues strongly to be associated with risk of HIV transmission but less so with risk of death from diseases other than HIV. A recent exception was a study by Kagaayi et al., which took place in rural Uganda and involved less intense follow-up. These investigators concluded that formula feeding should be discouraged in such a setting because of the resulting high-mortality rate from non-HIV causes.
Similar to most studies of HIV and infant feeding, Rollins et al. did not randomize mothers to feeding pattern. (Nduati et al. did so; Thior  and Kuhn et al. randomized to advice on when to stop breastfeeding.) With no randomization, there are two sources of potentially serious confounding acting in opposite directions. The first is that mothers who are at a more advanced stage of HIV disease are more likely to choose artificial feeding. In this study, CD4 cell counts were indeed lower for those who breastfed for a shorter period, though this was not significant.
The other risk of confounding is caused by the nearly universal finding that poorer mothers breastfeed for longer – and their children are more likely to die, whether they have HIV or not. Thus, it is unfortunate that a better proxy for poverty was not used in this study than maternal unemployment (apparently measured prenatally). The authors mention three factors maternal unemployment may be associated with: lack of ability to make independent infant feeding decisions, impaired nutrition, and lack of resources. However, as a proxy for poverty, maternal employment is heavily confounded in the South African context by its relationship to whether or not a woman is in a stable relationship with the child's father; the fact that mothers who work postnatally may do so either because they are so poor that they must work in spite of preferring to stay home with their baby or because career is important for them even though they may not need the money, at least soon after delivery; the fact that some working mothers will thus be richer than mothers who do not and others poorer (than women whose partners earn enough that they do not need to work); and the fact that infants of working mothers may be exposed to a caregiver with poorer understanding of or adherence to good hygiene practices than the mother or both.
We would certainly expect poverty to be associated with a lower HIV-free survival rate. As only about 10% of the sample avoided breastfeeding from birth, were these substantially wealthier than the rest? Were the 76% who continued breastfeeding beyond 6 months poorer than those who stopped earlier? If so, then these two feeding patterns will not result in similar rates of HIV-free survival when income is held constant.
The authors attempted to convince sample mothers that they should stop breastfeeding at 6 months, implying that all had access to nutritious foods for their infants. The assumption that doing so would lead to increased rates of HIV-free survival seems to be justified by their data. However, if poorer mothers are the ones who opt to breastfeed longer, we must be very careful in assuming that their infants will have the same outcomes as those who opt to breastfeed for shorter periods. This is especially true in South Africa where fears of stigma may be less of a factor explaining the continuation of breastfeeding beyond 6 months than poverty, at least compared with other African countries. Thus, WHO no longer recommends attempting to convince all HIV+ mothers to stop breastfeeding at 6 months: 'At 6 months, if replacement feeding is still not acceptable, feasible, affordable, sustainable and safe, continuation of breastfeeding with additional complementary foods is recommended, while the mother and baby continue to be regularly assessed (http://whqlibdoc.who.int/publications/2007/9789241595964_eng.pdf [accessed 12 Dec 2008]).
Finally, while both the article and editorial mention, respectively, that HAART treatment of eligible mothers and antiretroviral therapy (ART) prophylaxis are likely to reduce postnatal HIV transmission, the former is now increasingly available in Africa, making generalization from this study even more difficult.
1. Rollins NC, Becquet R, Bland RM, Coutsoudis A, Coovadia HM, Newell ML. Infant feeding, HIV transmission and mortality at 18 months: the need for appropriate choices by mothers and prioritization within programmes. AIDS 2008; 22:2349–2357.
2. Msellati P, Van de Perre P. Maternal HIV infection and infant feeding: support is the key. AIDS 2008; 22:2391–2392.
3. Kagaayi J, Gray RH, Brahmbhatt H, Kigozi G, Nalugoda F, Wabwire-Mangen F, et al
. Survival of infants born to HIV-positive mothers, by feeding modality, in Rakai, Uganda. PLoS ONE 2008; 3:e3877.
4. Nduati R, John G, Mbori-Ngacha D, Richardson B, Overbaugh J, Mwatha A, et al
. Effect of breastfeeding and formula feeding on transmission of HIV-1: a randomized clinical trial. JAMA 2000; 283:1167–1174.
5. Thior I, Lockman S, Smeaton LM, Shapiro RL, Wester C, Heymann SJ, et al
, Mashi Study Team. Breastfeeding plus infant zidovudine prophylaxis for 6 months vs formula feeding plus infant zidovudine for 1 month to reduce mother-to-child HIV transmission in Botswana: a randomized trial: the Mashi Study. JAMA 2006; 296:794–805.
6. Kuhn L, Aldrovandi GM, Sinkala M, Kankasa C, Semrau K, Mwiya M, et al
, Zambia exclusive breastfeeding study. Effects of early, abrupt weaning on HIV-free survival of children in Zambia. N Engl J Med 2008; 359:130–141. Erratum in: N Engl J Med