HIV-Infected Women Who Breast-Feed Face No Greater Risk of Mortality Than HIV-Infected Women Who Formula Feed, Study Finds

A Discussion With Shahin Lockman, M.D.

In this study summary, Shahin Lockman, M.D., of Brigham and Women's Hospital in Boston, Mass., summarizes her study on the effect of breast-feeding versus formula feeding on maternal HIV disease progression, mortality and micronutrient levels in a randomized trial in Botswana. After the summary, Dr. Lockman answers questions from the audience.

Shahin Lockman: About two million HIV-infected women per year deliver a baby. WHO [World Health Organization] feeding recommendations recommend replacement feeding when it's affordable, feasible, safe, et cetera. But the vast majority of women globally with HIV infection still breast-feed -- and, in that instance, exclusive breast-feeding is recommended. The question is: Does this have any impact on maternal health?

Shahin Lockman, M.D.
Shahin Lockman, M.D.

The first data to really look at this were published in 2001 by Dr. Ruth Nduati from Kenya. They found, in a randomized trial of breast versus formula feeding among HIV-infected women in the pre-HAART [highly active antiretroviral therapy] era, that mortality was threefold higher in the women who breast-fed. It was 11% versus 4%, which was quite alarming.

Since then, several observational studies have not confirmed that finding. In fact, some of them have shown that breast-feeding women have lower mortality and lower health risk. There are strong reasons to believe that women who are able to breast-feed, or choose to breast-feed, actually may be healthier to begin with, or may stay healthier. So those observational studies are potentially heavily confounded. That's the backdrop for our study.

We added an analysis of maternal health outcomes to a randomized, controlled trial of feeding interventions to prevent mother-to-child transmission [MTCT] called the Mashi Study, which was conducted in Botswana.1 It recruited 1,200 HIV-infected women between 2001 and 2003. Five hundred ninety-eight were randomized to breast-feeding for six months with infant AZT [zidovudine, Retrovir] prophylaxis to try to prevent mother-to-child transmission, and 602 were randomized to formula feeding.

The main MTCT outcomes have been presented and are not the object of this discussion. Rather, we followed women for up to five years postpartum. The median duration of follow-up in both of the feeding arms was four and a half years, essentially the same in both of the feeding arms. We looked at the primary endpoint -- which was time to decline of CD4 to less than 200, time to AIDS illness or time to death -- in both of the feeding arms. We observed that 34% of the women in the breast-feeding arm, compared with 28% of women in the formula-feeding arm, progressed to the primary endpoint of AIDS or death. This did not reach statistical significance: In univariate analysis the P value was .8, and for multivariate it was .07. It was a borderline trend that persisted after adjusting for other factors.

One other observation that we made was in a subset of women for whom we tested micronutrient levels and inflammatory markers, such as high sensitivity CRP [C-reactive protein], at six months postpartum. At the end of this feeding intervention period, the CRP was significantly higher in the breast-feeding women than in the formula-feeding women, which was intriguing and not something that's been looked at before.

Overall, my take on these data is that they're quite reassuring. This is the second randomized feeding trial, perhaps the last one that will look at different feeding methods, and it was able to avoid some of the issues of confounding. We did not find higher mortalities: 3% in both arms. We did not find higher mortality in the breast-feeding women within this population, which actually had access to HAART -- I should mention that's an important factor. But we did find an intriguing, unexplainable trend towards faster disease progression in the breast-feeding women, as well as a higher CRP at the end of the breast-feeding intervention period.

Audience Member #1: Did I hear you correctly? Women who were breast-feeding had higher mortality rates?

Shahin Lockman: No, the mortality was low in both arms. It was 3% in both arms. The rate of disease progression -- progression to AIDS or death -- as a composite endpoint trended towards being a little bit higher in the breast-feeding arm, but was not technically statistically significant. That was largely driven by a slightly more rapid decline in CD4 cell count.

Audience Member #1: You said you were surprised at that finding, right?

Shahin Lockman: Actually, I think we're reassured that the mortality rate wasn't different. I think what we were intrigued by was a higher C-reactive protein (the inflammatory marker) level. That was a little bit surprising, and that's not something that's been looked at before to our knowledge in HIV-infected women in relation to feeding.

Audience Member #1: Do you see any hypothetical protective effect or deleterious effect of HIV on breast-feeding?

Shahin Lockman: With the randomized data being, I think, easier to interpret and less subject to all these other complicated confounders: I think in a population of women that has access to HAART, there is a possibility that breast-feeding could potentially accelerate HIV disease progression. But it doesn't seem to affect major overall clinical outcomes, as long as these women are able to have CD4 monitoring and access to HAART, which this population did have.

Audience Member #2: You mentioned some concern that these data might be confused by the fact that women who chose to breast-feed tended to be more healthy. Were you able to control for that?

Shahin Lockman: Yes. There were only two randomized trials. This study was in 1,200 HIV-infected women who agreed to be randomized and actually had very, very high levels of adherence. When we looked at baseline characteristics, they were identical, and follow-up time was identical. That's what makes these data very powerful. The observational studies have other potential problems.

Audience Member #3: You said that you did not see a significant difference in the women who were eligible for HAART. Do you know about the women who were not? I also have a question about whether you had any finer measures of disease progression than just progression to death or to eligibility.

Shahin Lockman: For this breast-feeding population, other than 71 women who started antiretroviral treatment during pregnancy or before pregnancy, none of the others were on treatment. We looked at the time to progression -- to the need for treatment, essentially -- between the two arms. That's where we had this hint that maybe it was a little faster in the breast-feeding arm. But as soon as women needed treatment, they were able to get it. The overall mortality did not differ between the two arms.

We also looked at time to AIDS-defining illness using the CDC [U.S. Centers for Disease Control and Prevention] criteria, as well as CD4 decline and death. We're planning to look at the time to any diagnosis, in light of the finding of elevated CRP and the question of whether there may potentially be a link between inflammation and other non-AIDS-related outcomes. We haven't looked at that yet.

This transcript has been lightly edited for clarity.


  1. Lockman S, Ghebremichael M, Shapiro R, et al. The effect of breast feeding vs. formula feeding on maternal HIV disease progression, mortality, and micronutrient levels in a 1,200-person randomized trial, Botswana. In: Program and abstracts of the 16th Conference on Retroviruses and Opportunistic Infections; February 8-11, 2009; Montréal, Canada. Abstract 176.

Download the audio file