Hiv/Aids, Infant Feeding

1

“HIV/AIDS, Infant Feeding, and Human Rights,” in Wenche Barth Eide and Uwe Kracht, eds., Food and Human Rights in Development. Volume I. Legal and Institutional Dimensions and Selected Topics (Antwerp, Belgium: Intersentia, 2005), pp. 391-424.

Chapter 15

HIV/AIDS, INFANT FEEDING,

AND HUMAN RIGHTS

George Kent[*]

1Questionable coercion

Since the 1980s there has been a widespread belief that human immunodeficiency virus (HIV) can be transmitted from mother to infant through breastmilk (Hanson, 1988). One result has been that some women who have been diagnosed as being infected with HIV have been pressured into using breastmilk substitutes such as infant formula to feed their infants. Is that warranted?

In some cases this pressure has gone well beyond gentle urging by concerned health workers, and has taken the form of strong coercion by government agents. For example, in Los Angeles, a woman diagnosed as HIV-positive was confronted by social workers from the Child and Family Services agency because she was breastfeeding her child. They told her to go with them and have herself and her baby tested, or they would take the baby. On the way to the testing site the officials stopped to buy infant formula, and demanded that the woman stop breastfeeding immediately (Farber, 1998).

On September 17, 1998, Kathleen Tyson of Eugene, Oregon, then six months pregnant, was told that her blood tests indicated that she was HIV-positive. Her son, Felix, was born on December 7, 1998. He appeared to be healthy in every way. Less than 24 hours after his birth, Ms. Tyson was pressed by a paediatrician to treat Felix with AZT, an antiretroviral drug, and to not breastfeed him.

Having studied the issue along with her husband, she declined to accept that advice. Within hours, a petitioner from Juvenile Court came to her hospital room, and issued a summons for her to appear in court two days later. She and her husband were initially charged with “intent to harm” the baby, but the petition, dated December 10, 1998, said that the child “has been subjected to threat of harm.” When the Tysons appeared in court, they were ordered to begin administering AZT to Felix every six hours for six weeks, and to stop breastfeeding completely. The court took legal custody of the infant, but allowed the Tysons to retain physical custody so long as they obeyed the court’s orders. A hearing was held in Eugene from April 16 to April 20, 1999. The judge ruled against the Tysons. Thus, the State retained legal custody of Felix. The Tysons retained physical custody on the condition that, as ordered, Felix would not be breastfed (Tyson, 1999; Wolf, 2001).

I have a special interest in the Tyson case because I was asked to serve as an expert witness on its human rights aspects. However, after I was sworn in, the judge said that human rights considerations were irrelevant, and asked me to step down (Kent, 1999a; 1999b).

In another case, in Camden, England, a woman who had been diagnosed as HIV positive was ordered by the courts to have her infant tested for HIV. She and her partner refused, and they fled the country with the child, partly out of fear that if the child were to be tested and found to be HIV-negative, the mother might be ordered to stop breastfeeding (BBC News, 1999). They were fearful partly because they were aware of the outcome in the Tyson case in Oregon (Rosser, 1999; Corbett, 2000).

There have been other cases in which women diagnosed as HIV-positive have been forced to have their infants tested for HIV or to accept particular drug treatments. In this chapter, however, the focus is on situations in which the choice of infant feeding was a central issue.

In some countries such as Botswana, South Africa, and Thailand, free infant formula has been provided to mothers who have been diagnosed as HIV-positive. The provision of something of value at no cost can easily establish a kind of pressure on HIV-infected women to feed their infants with formula. In some cases, counselling programmes intended to be non-directive may end up as directive if they are pressing the client to make a particular choice. Some counsellors may take the view that a good HIV/AIDS counsellor is one who can convince pregnant women to do what the counsellor wants (de Paoli et al., 2002, p.147).

This chapter addresses difficulties in deciding how infants of mothers diagnosed as HIV-positive should be fed. The issues arise primarily out of scientific uncertainties regarding the likely impact of different feeding methods on the health of the infant. These uncertainties raise questions related to the infant’s human right to adequate food and human right to enjoyment of the highest attainable standard of health. They also raise questions about whether the mother’s right to make an informed choice should be suspended, and whether the State and others are obligated to provide specific kinds of information.

The chapter is organised as follows: Section 2 discusses the evidence for mother-to-child transmission of the HIV virus and the associated morbidity and mortality. Section 3 discusses the appropriateness of persuading HIV-positive mothers to feed their infants in a particular way, given the scientific uncertainties regarding the transmission of the virus through breastfeeding and also breastfeeding’s impact on the health of infants of HIV-positive mothers. Section 4 reviews current global policy recommendations for infant feeding by mothers diagnosed as HIV-positive. Section 5 examines human rights law and principles relating to infant feeding that are of special interest to the focus of the chapter. Section 6 then discusses whether the HIV/AIDS context is so exceptional that it warrants suspension of recommendations and principles that normally apply in non-HIV/AIDS contexts. Section 7 draws the conclusions, summing up central findings and principles relating to infant feeding when mothers are diagnosed as HIV-positive.

2 Uncertainty regarding HIV transmission through breastfeeding

How should mothers who are HIV-positive feed their infants? The major tension arises out of the fact that under some circumstances the dangers of using breastmilk substitutes might outweigh the dangers from being infected with HIV through breastfeeding. Counselling is supposed to support mothers in making informed choices among alternative feeding methods and safely carrying out those choices. However, the quality of counselling is necessarily limited by the quality of information available. As indicated in the following subsections, there are several different kinds of uncertainty relating to the transmission of HIV from mothers to infants.

2.1Definitions and indicators of transmission and infection

There is some disagreement regarding the very meaning of ’HIV infection‘, and there is also a great deal of variation in the ways in which HIV status is defined and measured. The complexity and the variations are illustrated by reports from the United States Centers for Disease Control (CDC) on the coding of HIV in children (CDC, 1994). There are many other definitions used in other places, they vary over time, and the level of understanding of health practitioners in the field is always questionable. If published research reports were more explicit about the rules by which they coded the HIV status of mothers and infants, they probably would show little consistency. This problem has been evident in the meta-analyses of studies on transmission of HIV from mother to infant (Dunn et al., 1992).

Though they cannot always be distinguished empirically, conceptually it is useful to distinguish among the sequential stages that are possible with any kind of infection: exposure, transmission, infection, disease, and death. The linkages at each stage usually occur with a time lag, and with less than one hundred percent certainty.

The terminology of the virus transmission discussion suggests that if even a small number of ‘copies’ of the virus pass from the mother to infant (transmission), that infant is thereby infected - by definition. However, some analysts emphasise the importance of ’seroconversion’, i.e. the change in the bloodstream that may result from the activity of the virus. Transmission is instantaneous, but seroconversion, or the process of infection, might take weeks or even months. The evidence of seroconversion is the presence of antibodies in the bloodstream. It is important to distinguish infection in the first sense (presence of the virus) from infection in the second sense (seroconversion) because the first does not always lead to the second.[1]

Thus, there is a distinction between transmission and infection, and there is a time lag between them. There is also a distinction between the onset of infection and the onset of the symptoms of disease. The official definitions of AIDS specify lists of AIDS-defining diseases. There is often a long time lag between the detection of HIV infection and the detection of any AIDS-defining disease. And there is another time lag between the onset of AIDS and subsequent AIDS-caused death, if it should occur. All of these time lags can make it difficult to track the chain of causation.

The rules for defining breastfeeding also vary a great deal, with some studies making little or no distinction between mixed feeding and exclusive breastfeeding, and some giving little attention to the duration of breastfeeding (Crowe et al., 2004). In most studies, samples were not randomised as to feeding methods. No study of mother to child transmission has drawn from a sample of the global population of mothers diagnosed as HIV-positive, so global generalisations are questionable. There are many confounding factors, so there is little reason to believe that a finding in, say, Kenya, would apply to, say, Oregon.

The introduction of various forms of drug treatment confuses matters even further. Recommendations for drug treatments are often accompanied by recommendations that the infants should not be breastfed. It is sometimes suggested that breastmilk substitutes should be used because of the drug treatment, as if the drug treatment somehow made breastfeeding more dangerous. Some drugs have been shown to be effective in reducing virus transmission regardless of whether or not the infant is breastfed (NIAIDNEWS, 2000). There is little clear evidence of interaction effects, either positive or negative, between the impacts of particular drug treatments and the impacts of particular feeding methods.

When infants are diagnosed as HIV-positive, it is difficult to determine whether the virus was transmitted during the pregnancy, the birth process, or breastfeeding. It is commonly assumed that increases in viral load in breastfed infants are attributable to the breastfeeding. However, there is evidence that the viral load in the infant can increase after birth even in the absence of breastfeeding. Consider Table 1 on the rates of virus transmission for breastfed and formula fed infants in several different studies.

In the studies in South Africa and Kenya, not only the breastfed infants but also the formula fed infants appear to have had increasingly high rates of infection over time. What does that mean? One possibility is that some mother-infant pairs were not categorised accurately. In the Kenya study, infants who were mixed feeders (mothers who used both breastfeeding and formula) were supposed to be placed with the breastfeeding cohort (Nduati et al., 2000; Mbori-Ngacha et al., 2001) However, while that was the intent, about 25% of the women who were assigned to the formula feeding group were reported to have also been giving breastmilk (Coovadia and Coutsoudis, 2001). New infections may have appeared after birth because some of the women who were supposed to be formula feeding were sometimes breastfeeding.

Study / Group / Rate of infant HIV infection (%)
At birth / 6 weeks / Months
3 / 6 / 15-18 / 24
South Africa / Breast-fed
(n = 394) / 6.9 / 19.9 / 21.8 / 24.2 / 31.6 / ---
Formula
(n = 157) / 7.6 / 18.0 / 18.7 / 19.4 / 19.4 / ---
Kenya / Breast-fed
(n = 191) / 7.0 / 19.9 / 24.5 / 28 / --- / 36.7
Formula
(n = 193) / 3.1 / 9.7 / 13.2 / 15.9 / 20.5
Brazil / Breast-fed
(n = 168) / --- / --- / --- / --- / 21 / ---
Formula
(n = 264) / --- / --- / --- / --- / 13 / ---

Table 1. Mother-to-Child Transmission Rates of HIV.

Source: Coovadia and Coutsoudis (2001)

Another possible explanation for the appearance of new infections after birth is that there might be a latency effect. The actual transmission of the virus may occur during pregnancy or delivery, but the biological activity that produces the antibodies may not rise to the level of detectability until some weeks or months after birth. If this is the case, the share of virus transmission attributable to breastfeeding might be much lower than has been supposed. According to this table, at six months the rate of HIV infection for breastfed infants in South Africa was only 4.8% higher than the rate for formula fed infants. Since there might well be a significant time lag between transmission of the virus and the onset of measurable infection, there is no way to know the exact moment of transmission, and thus there is no accurate way to determine which infections are due specifically to breastfeeding. It may be that in some cases the virus is transmitted prior to or during birth, but its effects do not become detectable until early in the infant’s life. Coovadia and Coutsoudis (2001, p.5) say

…the only data that we have currently available to estimate transmission in the first 6 months is to calculate the rate of new infections between 6 weeks and 6 months...

However, they do not make a clear distinction between transmission and infection, and do not acknowledge that there can be a significant time delay between the two.

2.2 Likelihood of transmission

There is considerable uncertainty about the likelihood of virus transmission through breastfeeding, and there is confusion about how that figure should be calculated (Dunn et al., 1992; Committee on Paediatric AIDS, 1995; Burr et al., 1997; Stoto et al., 1998; Crowe et al., 2004). Many of the reports about likelihood of transmission fail to say how their estimates were obtained. Many analysts have accepted Dunn’s estimate that the likelihood of transmission of the virus through breastfeeding is about 14%, but there were many flaws in his study (Crowe et al., 2004).

The likelihood of transmission of HIV from mother to infant through breastfeeding is different for different subpopulations and different kinds of circumstances (Recommendations, 1998, p.313). Thus, transmission risks can be meaningfully assessed only with regard to particular kinds of groups or conditions. In countries in which the use of drugs has sharply reduced the transmission rate, such as the United States, there are very few cases of perinatal HIV/AIDS. In the U.S. there were only 140 cases diagnosed in 2002 (CDC, 2004, Table 1).

Notably, there was a far higher incidence of perinatal HIV in the U.S. among Blacks than among Whites or Hispanics. Research should be undertaken to determine why Blacks have a high incidence of perinatal HIV despite their relatively low breastfeeding rate.

There is some indication that the likelihood of HIV infection may be lower if the infant is fed exclusively through breastfeeding, rather than in combination with other foods (Piwoz et al., 2004). It may be that feeding with anything but breastmilk risks damaging the gut of the infant, thus increasing the ease with which the HIV can pass into the child’s blood.

2.3 Morbidity and mortality

In 2000-2003, HIV/AIDS was estimated to cause about 321,000 child deaths worldwide per year, or about three percent of all deaths of children under five years of age (WHO, 2005, p.190). This large number of deaths does not mean that HIV-infection is always fatal. While many observers assume the worst, believing that infection means almost certain early death for the child (e.g., John et al., 2001; Bertolli et al., 2003), that has not been demonstrated to be the case.

What proportion of the children who contract HIV die early deaths? Unfortunately, the problem of paediatric HIV is hardly even mentioned in the 2004 Report on the Global AIDS Epidemic (UNAIDS, 2004). Another report says:

An estimated 800,000 children are newly infected with HIV annually. The majority of these children live in sub-Saharan Africa, where half of HIV-infected children die before their 5th birthday (Dabis et al., 2004, p.167)

The second sentence quoted here might suggest that if a child diagnosed as HIV-infected dies, it must have done so as a result of that infection. This may give more blame for child mortality to HIV infection than it deserves. It may be true that half of the HIV-infected children died before their fifth birthday, but many of them probably died from other causes. Walker et al. (2002) estimate that in sub-Saharan Africa in 1999, HIV/AIDS accounted for about 7.7% of under-5 deaths. In some countries the proportion was much lower, and in others much higher, with the highest estimated mean HIV-attributable under-5 mortality rate being 57.7%, for Botswana.

In a study in Rwanda, the estimated risk of death among infected children at 2 and 5 years of age was found to be 45% and 62% respectively (Spira et al., 1999). Another study indicated that about 22% of the HIV-positive children studied progressed to AIDS or death by ten years of age (European Collaborative Study, 2002). A review by the HIV Paediatric Prognostic Markers Collaborative Study Group (2003) found that 20-25% of children infected with HIV-1 progressed rapidly to AIDS or died during infancy, as shown by early cohort studies. In New York City the mortality rate for children diagnosed as HIV-positive is reported to be less than five percent. The city’s Department of Health and Mental Hygiene reported, in 2003: