The Risks of Not Breastfeeding

In developing countries, breast-feeding is both the cornerstone of child survival and the cause of about one-third of all infant HIV infections. Moreover, the same economic and development inequities that make breast-feeding so critical to infant survival in these settings also make formula feeding inaccessible, unfeasible, unaffordable, unsustainable, and unsafe (ie, not AFASS) for most families. This poignant dilemma has resulted in emotive and sometimes polarizing debate within the public health community as we have wrestled to quantify these competing risks, test interventions to reduce them, and modify policy as our understanding improves.

Four articles in this issue of the Journalof Acquired Immune Deficiency Syndromes shed new light on this issue.^1-4 To appreciate their design and findings, it is helpful to contextualize them within the rapidly evolving science and policy of HIV and infant feeding.

HIV AND INFANT FEEDING: 1985-2000

Following the first reported case of breast-feeding-associated HIV transmission in 1985,⁵ Joint United Nations Programme on HIV/AIDS, United Nations Children's Fund, and World Health Organization (WHO) jointly published divergent infant feeding guidance for HIV-infected women in developing and developed countries in 1987⁶ and 1992⁷ recommending that “where the primary causes of infant deaths are infectious diseases and malnutrition” (ie, developing countries) all women should breast-feed-including those known to be HIV positive, but “where infectious diseases are not the primary cause of infant death” (ie, developed countries) HIV-positive women should avoid all breast-feeding. Between 1994 and 1997, data from 4 cohort studies of HIV-exposed breast-fed and non-breast-fed infants estimated that breast-feeding resulted in a 4%-22% excess transmission risk.^8-11 As a result, United Nations' (UN) agencies issued revised guidance in 1998, acknowledging the substantial HIV infection risk associated with breast-feeding and recommending that “when children born to women living with HIV can be ensured uninterrupted access to a nutritionally adequate breast milk substitutes that are safely prepared and fed to them, they are at less risk of illness and death if they are not breast-fed.”¹² In 2000, publication of findings from Kenya of the first randomized trial of breast versus formula feeding affirmed the 1998 UN guidance: HIV-infection-free survival at 24 months was significantly higher among HIV-exposed infants randomized to formula feeding compared with breast-feeding (70% versus 58%).¹³ Also published in 2000 was a WHO-commissioned pooled analysis of data from 6 developing countries on the effect of breast-feeding on mortality.¹⁴ This analysis demonstrated that the odds of infant death due to not breast-feeding rapidly declines with age, especially after 6 months: 5.8 [95% confidence interval (CI): 3.4-9.8] among infants under 2 months of age; 2.6 (95% CI: 1.6-3.9) among infants aged 4-5 months; 1.8 (95% CI: 1.2-2.8) among infants 6-8 months of age; 1.4 (95% CI: 0.8-2.6) among infants 9-11 months of age; and 1.6 (95% CI: 0.8-3.2) to 2.1 (95% CI: 1.1-4.0) during the second year of life. Finally, in 1999, the first evidence that exclusive breast-feeding carries a lower risk of postnatal transmission compared with mixed feeding was published.¹⁵ Accordingly, when the joint UN agencies updated HIV and infant feeding policy in October 2000, the authors reiterated the previous recommendation that HIV-infected mothers avoid all breast-feeding when an AFASS replacement feeding was available, but added “otherwise, exclusive breast-feeding is recommended during the first months of life.” The committee also added a specific recommendation that mothers who lack an AFASS alternative and initiate breast-feeding should discontinue breast-feeding “as soon as feasible.” It was hoped that this guidance would provide an optimal compromise, maximizing breast-feeding benefit while minimizing HIV transmission risk. The guidance did not specify an age for breast-feeding cessation; stated that the local circumstances, the individual woman's situation, and the risks of replacement feeding should be considered; and urged that specific guidance be given to mothers who replacement feed. However, this new guidance was widely interpreted as recommending that all HIV-positive mothers stop breast-feeding by 6 months and implementation of this policy was not always accompanied by effective nutrition counseling.

Data reported in the 4 articles in this issue were collected when this “early breast-feeding cessation policy” was in place:

• Onyango-Makumbi et al⁴ (Kampala, Uganda) compare rates of serious gastroenteritis (GE) and mortality among HIV-exposed but uninfected infants enrolled in the HIV Network for Prevention Trials (HIVNET) 012 and HIV Immunoglobulin (HIVIGLOB) studies conducted before and after, respectively, early breast-feeding cessation policy was in place. Importantly, in both this study and the next one by Kafulafula et al, infants who became HIV-infected were censored from the analysis. Median duration of breast-feeding was 4.0 months in HIVNET 012 and 9.3 months in HIVIGLOB. Rates of serious GE were significantly higher in HIVIGLOB compared with HIVNET 012 (8.0 versus 3.1/1000 child-months [c-m]), and overall mortality was 50% higher, although did not reach statistical significance (3.2 versus 2.0 deaths/1000 c-m).
• Kafulafula et al³ (Malawi) compare GE and mortality rates among exposed but uninfected infants enrolled in the Nevirapine/AZT studies (NVAZ) and Post Exposure Prophylaxis for Infants (PEPI) studies, conducted before and after, respectively, early breast-feeding cessation policy was in place. At 9 months, 89% of NVAZ infants and 10% of PEPI infants were still breast-feeding. Rates of GE hospitalization between 7 and 12 months (2.28% versus 0.13%), GE mortality by 12 months (24 versus 12 deaths/1000 c-m), and total infant mortality (79 versus 67 deaths/1000 c-m) were all significantly higher among PEPI compared with NVAZ infants.
• Homsy et al² (Tororo, Uganda) report findings from a cohort of 102 infants born to HIV-positive women receiving highly active antiretroviral therapy (HAART) treatment. Mothers were advised to stop breast-feeding between 3 and 6 months; median breast-feeding duration was 5 months. By 18 months, none of the infants had tested HIV positive but 23 (19%) had died. Mortality was 6.2 (95% CI: 1.4-27.0) times higher among infants breast-fed less than 6 months compared with longer than 6 months.
• Creek et al¹ (Botswana) present investigation findings of a diarrhea outbreak that occurred in Botswana in early 2006 when free formula was being provided to HIV-positive women and early breast-feeding cessation was common among HIV-negative women. Among 153 children hospitalized for diarrhea, 88% were not breast-feeding; 64% and 18% were HIV exposed and infected, respectively. Of the 33 of the children (22%) who died, only 1 was breast-fed (a 1-month old who was also receiving cow milk and formula). Therefore, HIV infection was not associated with death.

Major Implications of These Articles

• These articles demonstrate that in these settings, provision of feeding interventions for HIV-exposed infants focused only during the first few months of life is not long enough; breast-feeding continues to provide substantial protective benefits against GE and mortality into the second year of life.
• Interventions provided by the studies to reduce risks associated with replacement feeding may have reduced them, but did not prevent them. These included infant cotramoxazole, point-of-use water treatment, free formula and complementary food, intensive nutrition education, and free clinical care. One intervention not offered by any of these studies was a toilet. Poor sanitation may cause chronic tropical enteropathy, which may be a major cause of undernutrition and poor health for all infants¹⁶ but especially for those not breast-fed.¹⁷ Toilets may be particularly important for HIV-exposed non-breast-fed infants; in the MASHI (meaning “milk” in Setswana) study, only 3 factors were retained in a multivariate model, explaining 24-month mortality of HIV-exposed infants who were not breast-fed after 6 months: infant HIV status, birth weight, and whether the family had a latrine or not.¹⁸ Sadly, Africa, where most HIV-exposed children live, is also the only region in the world projected to have substantially more people without adequate sanitation in 2015 than in 2005.¹⁹ Ventilated lined pit latrines provide excellent sanitation for a family for up to 20 years (while also minimizing odor and flies) and can be built for <US $50.
• The Homsy article,² together with other recently presented data from the Breastfeeding, Antiretroviral, and Nutrition Study (BAN),²⁰ Kesho Bora,²¹ and Mma Bara (meaning “mother baby” in Setswana),²² studies, presents the very good news that maternal HAART, whether given for maternal treatment or postnatal transmission prophylaxis, reduces breast-feeding-associated transmission to low rates. Furthermore, based on results from Mitra,²³ six-week extended dose nevirapine (SWEN),²⁴ BAN,²⁰ and PEPI,²⁵ antiretroviral prophylaxis given to the breast-feeding infant provides comparable protection. Preliminary data from these studies also suggest that these regimens are safe. How long these drug interventions could or should be provided to infants or to women who are not on HAART for their own health remains to be determined based on risk to benefit and cost to benefit analyses. In this regard, it is relevant to note that the protective effect of breast-feeding on mortality in the WHO pooled analysis was similar for children 6-12 months and 12-24 months of age, suggesting that these interventions are likely to provide benefit (in addition to possible risk and certain cost) for the first 2 years of life. Another alternative that may extend safer breast-feeding after discontinuing drug prophylaxis is expressing and heat treating breast milk.²⁶ This intervention may be easier to implement after 6 months when lactation is well established, infants can drink from a cup more easily, and milk comprises proportionately less of the infant's total diet.
• In addition to the short-term harm reported in these articles, early breast-feeding cessation may also engender longer term adverse effects. In the Zambian Exclusive Breastfeeding Study (ZEBS) study, which randomized infants at 4 months to continued breast-feeding or abrupt breast-feeding cessation, early cessation was associated with ∼ a 0.25 lower mean weight-for-age Z-score among 9-15 month infants,²⁷ a difference associated with about a 10% greater under-5-year mortality rate in non-HIV-exposed infants.²⁸ This risk may be even greater among HIV-exposed infants if the effects of growth faltering and HIV exposure (eg, sick or dead parents) are synergistic.
• There is one crucial issue not addressed in these reports: none present sufficient data to assess the balance of risks; all demonstrate that the risks of illness and death associated with not breast-feeding are high but none assess whether these rates exceed the HIV transmission rates that would have occurred had the mothers in HIVIGLOB, PEPI, and Tororo Uganda continued breast-feeding or had HIV-positive mothers in Botswana been breast-feeding during the diarrhea outbreak. Postnatal transmission data are not yet available for HIVIGLOB⁴ but have been published for NVAZ ²⁹ and PEPI.²⁵ Estimating this balance of risks must be circumspect because the cohorts were not concurrent and many differences could have influenced infection-free survival besides the drug intervention and breast-feeding duration. Nonetheless, although postnatal transmission was lower in PEPI than in NVAZ (3.5% versus 5.7% at 9 months³ and ∼8%²⁵ versus 9.68%²⁹ at 24 months), infection-free survival at 24 months among infants who were HIV polymerase chain reaction negative at 6 weeks was lower: ∼83% in all 3 PEPI arms²⁵ (estimating from graphs in the article) compared with 87.4%²⁹ in the NVAZ studies, suggesting that early breast-feeding cessation may have resulted in net harm.

2006 HIV AND INFANT FEEDING POLICY

In October 2006, when HIV and infant feeding policy was again revised; new evidence included an additional evidence that early exclusive breast-feeding is protective against postnatal transmission compared with mixed feeding,^30,31 findings from 2 randomized trials in which infection-free survival was comparable between infants randomized to formula or breast-feeding from birth (MASHI³²) and between infants randomized at 4 months of age to continued breast-feeding or abrupt breast-feeding cessation (ZEBS³³); and preliminary findings from the 4 studies published in this issue of the Journal of Acquired Immune Deficiency Syndromes. Reflecting these new data, the 2000 policy which had stated, “When replacement feeding is AFASS, avoidance of all breastfeeding by HIV-infected mothers is recommended… Otherwise, exclusive breastfeeding is recommended during the first months of life,” was rewritten to read “Exclusive breastfeeding is recommended for HIV-infected women for the first six months of life unless replacement feeding is AFASS for them…. When replacement feeding is AFASS, avoidance of all breastfeeding by HIV-infected women is recommended.” The 2006 guidance also specified that breast-feeding should continue beyond 6 months if a replacement feeding is still not AFASS. Thus, even though replacement feeding remains the best option for HIV-infected women in situations with plentiful nutritious foods and breastmilk substitutes, good sanitation, clean water, cooking fuel, accessible quality health care delivered by a well-functioning health care system, high levels of female education, low rates of underlying infant mortality, and broad social support; and even though there are examples of successful replacement feeding in Africa,^34-36 this subtle shift (in nuance more than substance) in the recommendations reflected the participants' sentiments that:

• For the large majority of HIV-exposed infants, avoidance of breast-feeding will not improve infection-free survival and may reduce it.
• For the large majority of HIV-exposed infants, exclusive breast-feeding for the first 6 months of life, and continued breast-feeding after that for some yet undetermined duration, will provide the greatest chance of infection-free survival.

Ironically, the HIV epidemic may be the best thing that ever happened to breast-feeding. It is difficult to imagine an experiment that could provide more compelling momentum for breast-feeding promotion than the natural one the HIV epidemic has provided: an incurable disease that infects up to 30%-40% of antenatal women in some African countries and is transmitted to infants through breast-feeding leads UN agencies and governments to promote and even freely provide formula for exposed infants. Moreover, scores of these infants are already under surveillance by scientists measuring and documenting numerous indices of infant health. In short, the HIV epidemic and our efforts to ameliorate its effect on children provided an ethical opportunity to observe what happens when large number of infants living in conditions of poverty are not breast-fed. If these observations lead to stronger breast-feeding policy and programming that in turn reduce the 1.4 million child deaths occurring each year due to suboptimal breast-feeding,³⁷ we will have created one of the epidemic's very few silver linings.