The HIV-exposed, uninfected African child

Authors


Corresponding Author S. Filteau, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, UK. E-mail: suzanne.filteau@lshtm.ac.uk

Summary

The increasing success of prevention of mother-to-child HIV transmission programmes means that in Africa, very large numbers of HIV-exposed, uninfected (HIV-EU) children are being born. Any health problems that these children may have will thus be of enormous public health importance, but to date have been largely neglected. There is some evidence that HIV-EU African children are at increased risk of mortality, morbidity and slower early growth than their HIV-unexposed counterparts. A likely major cause of this impaired health is less exposure to breast milk as mothers are either less able to breastfeed or stop breastfeeding early to protect their infant from HIV infection. Other contributing factors are parental illness or death resulting in reduced care of the children, increased exposure to other infections and possibly exposure to antiretroviral drugs. A broad approach for psychosocial support of HIV-affected families is needed to improve health of HIV-EU children. High quality programmatic research is needed to determine how to deliver such care.

Abstract

Le succès croissant des programmes de prévention de la transmission mère-enfant du VIH signifie qu’en Afrique naissent maintenant un très grand nombre d’enfants exposés au VIH mais non infectés. Tout problème de santé que ces enfants pourraient avoir sera donc d’une grande importance pour la santé publique, mais à ce jour aura été largement négligé. Il existe des preuves que les enfants africains exposés au VIH mais non infectés sont exposés à un risque accru de mortalité, de morbidité et de ralentissement de la croissance initiale par rapport à ceux non exposés. Une cause principale probable de cette altération de la santé est l’exposition réduite au lait maternel comme les mères sont moins en mesure d’allaiter ou arrêtent l’allaitement maternel précocement afin de protéger leur bébé contre l’infection par le VIH. D’autres facteurs contribuants sont la maladie des parents ou leur décès résultant en une réduction des soins pour les enfants, l’augmentation de l’exposition à d’autres infections et éventuellement, l’exposition aux médicaments antirétroviraux. Une approche globale pour le soutien psychosocial des familles touchées par le VIH est nécessaire pour améliorer la santé des enfants exposés au VIH mais non infectés. Une recherche programmatique de haute qualité est nécessaire pour déterminer comment délivrer ces soins.

Abstract

El éxito en aumento de los programas de prevención de la transmisión vertical del VIH significa que en África están naciendo un gran número de niños expuestos al VIH, pero no infectados (VIH-ENI). Por lo tanto, cualquier problema de salud que tengan estos niños tendrá una enorme importancia desde un punto de vista de salud pública, y sin embargo hasta la fecha no han sido tenidos en cuenta. Hay alguna evidencia de que los niños africanos VIH-ENI tienen un mayor riesgo de mortalidad, morbilidad y un crecimiento más lento que sus contrapartes no expuestas al VIH. Una posible causa del deterioro en la salud de estos niños es su menor exposición a la leche materna, puesto que sus madres son o menos capaces de lactar o deben dejar de dar el pecho antes de tiempo para proteger al niño de la infección por VIH. Otros factores que contribuyen son la enfermedad o muerte de los padres – que resulta en un menor nivel de cuidado de los niños y una mayor exposición a otras infecciones, así como la posible exposición a medicamentos antirretrovirales. Se requiere de una aproximación amplia de apoyo psicosocial a las familias afectadas por el VIH con el fin de mejorar la salud de los niños VIH-ENI. Una investigación programática de calidad es necesaria para determinar las vías de entrega de dichos cuidados.

Introduction

HIV infection of infants by mother-to-child transmission (MTCT) is one of the most emotive and tragic aspects of the HIV epidemic. In the past decade, there has been considerable good news on that front. In industrialised countries, MTCT has decreased to negligible levels through a combination of highly active antiretroviral therapy (HAART), elective caesarean section and avoidance of breastfeeding (Newell et al. 2004). In Sub-Saharan Africa, the roll-out of both HAART and more limited perinatal antiretroviral therapy (ART), as well as improved understanding of risk factors for transmission through breastfeeding, have dramatically reduced the number of infants becoming infected.

This lower rate of MTCT means that, in spite of decreasing antenatal HIV prevalence in some countries (UNAIDS 2008), the number of HIV-1-exposed, uninfected (HIV-EU) children being born is rising and dwarfing even the staggering number of children infected with HIV. Several countries in sub-Saharan Africa have antenatal HIV infection prevalences of more than 20% (UNAIDS 2008). If MTCT is reduced to 10% or less through a combination of ART and short duration of exclusive breastfeeding (Kumwenda et al. 2008), this means that more than 18% of all infants born in these countries are HIV-EU. Any health problems that these children may have will thus be of enormous public health importance. Furthermore, these children are crucial to their countries’ economic and social future, not only because of their own work and other contributions to their societies but also because they are likely to shoulder much of the burden of care of their HIV-infected relatives.

The evidence regarding health of HIV-EU children is disparate and there appear to be no published intervention trials specifically addressing their health. Hence, the information does not lend itself to systematic review. This review aims to collate the evidence for health problems among HIV-EU African children, to suggest potential causes of the problems and to identify research gaps and health programme needs.

Limitations of the available information

Most available information comes from studies in industrialised countries, which may not be applicable to the African child for several reasons. First, unlike in Africa where the HIV epidemic is generalised across the socioeconomic spectrum (De Cock et al. 2002), most HIV-exposed children in industrialised countries come from marginalised sectors of society. Results from industrialised countries may be confounded by parental drug abuse or particularly stressful home environments. Second, the underlying exposures to diet and infection differ markedly between Africa and Europe or North America. Poor diets and malnutrition are common even among HIV-unexposed children in Africa; as are exposure to malaria and other infections. Third, because of poverty, unavailability of appropriate replacement foods and high risk of infectious disease morbidity and mortality, the World Health Organization (WHO) advises that most HIV-infected African women breastfeed their infants (WHO Briefing note: HIV and Infant Feeding 2007). This is in contrast to industrialised countries with readily available infant formula, lower risk of infections and better access to health care. Whether or not an infant is breastfed will have major effects on growth (Dewey 1998), immune system development (Hanson 2000) and exposure to HIV.

Although much research has been conducted with HIV-infected African women and their infants, virtually all of it has focussed on prevention of MTCT through ART or different infant feeding modes. In most cases, HIV-uninfected women were not included. Without control HIV-unexposed children, it is impossible to determine whether high risk of mortality or growth faltering in HIV-EU children results from the HIV or from other aspects of poverty. Because of the broad similarity of HIV-affected and unaffected households in Africa, children in these households can be readily compared.

Evidence for health problems of HIV-exposed, uninfected children

Mortality

Mortality among HIV-EU African children was first examined in MTCT trials, which included only HIV-infected mothers. A pooled analysis of seven such trials (Newell et al. 2004) found that risk factors for death among HIV-EU infants and young children were maternal death and lower maternal CD4 count. In a further prevention of MTCT trial not included in that review, low birth-weight was a risk factor for death among HIV-EU Tanzanian infants (Wei et al. 2004). A recent large MTCT trial in Zimbabwe (Marinda et al. 2007) included HIV-uninfected women and found that mortality within the first 2 years was significantly higher among HIV-EU children than among HIV-unexposed children (Figure 1). Risk factors for death among HIV-EU children were lower birth weight, male sex, maternal death, low maternal CD4 count, severe maternal anaemia, single or widowed mother and low household income.

Figure 1.

 Mortality among HIV-exposed, uninfected and unexposed African children. Data are percentages of deaths among Botswanan children to age 24 months (Shapiro et al. 2007), Ugandan children to age 24 months (Brahmbhatt et al. 2006), Gambian children to age 5 years (Van der Loeff et al. 2003), Malawian children between ages 12 and 36 months (Taha et al. 1999), Zambian children between ages 9 and 36 months (Sutcliffe et al. 2008) and Zimbabwean children to age 24 months (Marinda et al. 2007). Numbers over the bars refer to the total number of children (the denominator for the mortality percent) in each group. P-values for comparison of percent mortality between HIV-exposed, uninfected and unexposed children are: Botswana, P = 0.002; Uganda, not significant at 24 months, P < 0.05 at 18 months; Gambia, P = 0.17; Malawi, not significant, P-value not provided; Zambia, P = 0.13; Zimbabwe, P < 0.05.

Other mortality information has come from longitudinal community surveillance. A pooled analysis of three such studies from Uganda, Tanzania and Malawi found that both maternal HIV infection and maternal death increased the risk of child death; child HIV status information was unavailable (Zaba et al. 2005). Several cohort studies distinguished mortality among HIV-EU children from among HIV-infected ones (Figure 1). In Botswana, mortality was significantly higher among HIV-EU children compared with HIV-unexposed children with discontinuation of breastfeeding the strongest predictor of illness (Shapiro et al. 2007). In Uganda (Brahmbhatt et al. 2006), cumulative mortality rate at 12, 18 and 24 months tended to be higher, statistically significantly so only at 18 months, among HIV-EU children than among HIV-unexposed children. In Gambia, under-5 mortality hazard of HIV-EU was not significantly higher than that of HIV-unexposed children; maternal death was a major risk factor for child death, irrespective of maternal or child HIV status (Van der Loeff et al. 2003). A study in Malawi, which analysed mortality between ages 12 and 36 months, found no significant difference between HIV-EU and unexposed children (Taha et al. 1999). In Zambia, mortality between 9 and 36 months of age among HIV-EU children was slightly, but not significantly, higher than among HIV-unexposed children (Sutcliffe et al. 2008). In Rwanda (Spira et al. 1999), there was no difference in child mortality between HIV-EU and unexposed children (exact data were not presented).

Although data are limited, it seems that the mortality of HIV-EU children is generally greater than that of HIV-unexposed children. In addition to relatively small sample sizes and numbers of deaths, it is likely that both underlying child mortality rate and causes in the area and availability of medical care contribute to differences among studies.

Growth

Growth is a sensitive, if non-specific, indicator of overall infant and child health, which has been compared in several cohorts between HIV-EU and HIV-unexposed infants. Results may be affected by the fact that infants of HIV-infected mothers often have lower birth weight than infants of uninfected mothers, even when the infants themselves are not infected (Bailey et al. 1999; Marinda et al. 2007).

Growth of HIV-EU children in the European Collaborative Study was not lower and possibly higher for weight and body mass index, although that may reflect a secular trend, than children in the 1990 British Growth Standards (European Collaborative Study 2003a). In the Democratic Republic of Congo (DRC), although there was no long-term effect of maternal HIV status on growth of HIV-uninfected children, the HIV-EU infants had lower lengths, but not weights, than the unexposed infants for the first half of infancy (Bailey et al. 1999). In Rwanda, although again there were no overall effects of maternal HIV status over the first 4 years of life, HIV-EU infants had statistically lower length through most of infancy and lower weight at 6 months (Lepage et al. 1996). In Zambian infants followed for the first 16 weeks of life, both weight and length adjusted for gestational age, tended to be lower among HIV-EU compared with unexposed infants, although significantly so only for weight at 6 weeks (Makasa et al. 2007). The differences in growth between HIV-EU and unexposed infants, although often statistically significant, are generally small (about 0.2 to 0.3 Z scores) and transient. It is unclear whether these differences have adverse effects in the short term or whether, occurring at a critical point in the infant’s development, they predispose the child for subsequent poorer health, obesity, chronic disease or cognitive dysfunction (Singhal & Lucas 2004; Gillman 2008).

Compared to European cohorts, HIV-EU African children show less rapid catch up growth after being born small. This likely reflects the fact that HIV-exposed infants are generally formula-fed in Europe, but breastfed in Africa. Breastfeeding may affect health and cognitive development in addition to growth (Horta et al. 2007); therefore, aiming for catch-up growth in itself, without attention to other aspects of infant health, would be unwise.

Morbidity

Infections may be more common and severe among HIV-EU children than among unexposed children, which could contribute to both slower growth and higher mortality. An early study in Zaire (DRC) found that the incidence of persistent diarrhoea (>14 days) among HIV-EU infants was twice that among HIV-unexposed infants (Thea et al. 1993). Maternal death was a risk factor for persistent diarrhoea, but other factors, including birth weight, gestational age at birth and age at weaning, were not. HIV-EU Kenyan children hospitalised for acute malaria were more likely to develop the complication of severe anaemia than were HIV-unexposed children (Otieno et al. 2006).

There have been reports of pneumonia caused by Pneumocystis jiroveci, normally considered an opportunistic pathogen in HIV infection, among HIV-EU infants (Heresi et al. 1997; McNally et al. 2007). In South Africa, HIV-EU infants hospitalised with pneumonia tended to have higher treatment failure rates than HIV-unexposed infants; P. jiroveci was one of the most common pathogens in both HIV-infected and HIV-EU infants who failed first-line treatment, but was not found among HIV-unexposed infants who were treatment-failures (McNally et al. 2007).

Cognitive and motor development

Although HIV-infected children show delays in motor and possibly language, development (Abubakar et al. 2008), there is little adequately controlled information concerning development of HIV-EU children, especially in Africa. In Uganda, HIV exposure did not affect neurological status, information processing or mental and motor development as measured by Bayley Scales of Infant Development (Drotar et al. 1997). Analysis of a subgroup of these children followed up at about age 9 years also did not show any developmental differences between HIV-EU and HIV-unexposed children (Bagenda et al. 2006). The lack of difference in home environment and mother-child interactions between HIV-infected and uninfected mothers likely contributed to the similar development of their children (Drotar et al. 1997). In Rwanda, using a non-standard tool designed to assess several aspects of neurological and social development, HIV-EU children did not differ from HIV-unexposed children in gross motor, fine motor, language or social contact (Msellati et al. 1993). In the DRC, using a variety of tests depending on the children’s age, HIV-EU children, aged 18–72 months, had poorer motor, but not mental, development and delays in language expression than HIV-unexposed children (Van Rie et al. 2008). It is likely important that this study specifically selected HIV-EU children living with mothers with symptomatic HIV and their families were poorer than the families of HIV-unexposed children; these family and socioeconomic differences, not present in the Ugandan (Drotar et al. 1997) or Rwandan (Msellati et al. 1993) studies, may have contributed to developmental delays. Overall, it seems that most detectable developmental delays among HIV-EU children are a result of the environment, not of any direct biological consequences of being exposed to HIV. However, it is interesting that maternal multivitamin supplements protected Tanzanian infants against motor developmental delay (McGrath et al. 2006).

Potential causes of poor health and nutrition among HIV-EU children

Lack of parental care

Maternal death and orphanhood are well-established risks factors for poor infant and child health and survival (Miller et al. 2007; Watts et al. 2007). Studies of mortality cited above have generally found that maternal death is a risk for infant and young child death, irrespective of either maternal or infant HIV status. Maternal health does not need to be so poor as to lead to death for an HIV-EU child to be adversely affected; low CD4 count (Kuhn et al. 2005; Marinda et al. 2007), maternal anaemia (Chatterjee et al. 2007) and low birth-weight (Wei et al. 2004), which could reflect poor maternal health during pregnancy, also increased the risk of infant death. These results emphasise the importance of primary prevention of maternal HIV infection and of maintaining health of HIV-infected mothers. The roll out of ART and increased access to prophylactic antibiotics should contribute to improved parental and thus child survival (Walter et al. 2006).

However, orphanhood does not necessarily result in poor health, poverty or neglect (Linblade et al. 2003; Richter 2006; Parikh et al. 2007). The extended family social structure in much of Africa appears to be able to absorb at least some orphans and protect their health and opportunities for development. The UNICEF model of causes of malnutrition (UNICEF 1998) recognises the important contribution of care to child health. Care has many components, which are difficult to measure in large trials or community surveys. A study in Kenya found no difference in a number of care practices between HIV-infected and uninfected women (Sherry et al. 2000). There is, however, evidence for the effects of one key component of care – infant feeding – in studies of HIV-EU children (see below). In addition, there is evidence in some situations that HIV-affected households may be poorer than HIV-unaffected households in the area because of a combination of decreased ability to work and earn by adults and increased expenditure on health care.

Infant feeding practices

The most important modifiable factor affecting health of HIV-EU children is infant feeding practice; the subject has been recently reviewed for this journal (Coovadia & Bland 2007) and will not be discussed in detail here. Note that strong and articulate contrary views are also held (Farmer et al. 2006). Reduced breastfeeding by HIV-infected mothers could account for increased morbidity and mortality, increased exposure to dietary pathogens, altered immune functions and growth and potentially slower development of HIV-EU infants compared to unexposed infants. WHO summarises the results of much research in their recommendation that ‘Exclusive breastfeeding is recommended during the first months of life unless replacement feeding is acceptable, feasible, affordable, sustainable and safe’ (AFASS; WHO 2007). Given not only the many benefits of breastfeeding but also the frequent lack of AFASS alternatives (Becquet et al. 2006; Lunney et al. 2008; Owino et al. 2008), early exclusive breastfeeding is the safest option for both infected and uninfected infants of HIV-infected mothers (Coovadia & Bland 2007). Unfortunately, confusion about infant feeding messages may lead HIV-infected women to stop breastfeeding early even when they lack AFASS alternatives (Doherty et al. 2007).

Less breastfeeding may explain some of the association between sick or dead mothers and higher risk of infant mortality. In Guinea-Bissau, where HIV prevalence was low, maternal death was a higher risk for death of infants than for older children, which the authors suggested was because of the deprivation of maternal breast milk (Masmas et al. 2004). Even surviving mothers, if ill, may not breastfeed optimally. The median time of stopping exclusive breastfeeding in a group of Zambian mothers was 6 weeks in spite of strong encouragement and support for exclusive breastfeeding (Chisenga et al. 2005); the major risk factors for early cessation were primiparity and maternal systemic illness, the latter more common among HIV-infected women. Although HIV-infected South African mothers who were strongly supported to breastfeed exclusively did not experience significantly more clinical mastitis than uninfected women (Bland et al. 2007), similarly supported HIV-infected Zambian mothers were more likely to exhibit subclinical mastitis (Kasonka et al. 2006). Subclinical mastitis in dairy cattle is associated with reduced milk production (Shuster et al. 1995). It would be of interest to determine if subclinical mastitis lowers breast milk output even among women exclusively breastfeeding as subclinical mastitis is associated with reduced growth of infants of HIV-infected and uninfected women (Filteau et al. 1999; Gomo et al. 2003; Kasonka et al. 2006).

It is a concern that reductions, recommended and observed, in breastfeeding by HIV-infected mothers are not generally being accompanied by compensatory improvements in AFASS replacement foods. Among HIV-infected women in Cote d’Ivoire, a purposively designed infant feeding score was low at 6 months because of limited dietary diversity and food frequency; low score at 6 months was associated with subsequent risk of stunting (Becquet et al. 2006). Among HIV-infected Zimbabwean women, lack of availability of replacement foods meant that diets of non-breastfed infants were insufficient in energy and most micronutrients (Lunney et al. 2008). Zambian mothers were knowledgeable about optimal infant feeding, but actual practices were constrained by cost of high quality foods (Owino et al. 2008). Diets of young HIV-EU South African were less diverse, although not significantly so, than diets of HIV-unexposed children (Mpontshane et al. 2008). The study did not investigate reasons for lower dietary diversity, which can result in poorer nutritional status and health, but both economic and maternal care factors could have contributed.

The transition period between exclusive breastfeeding and complete cessation of breastfeeding has been a concern. The association between exclusive breastfeeding and lower MTCT has led some to interpret the relationship as causal. If mixed feeding has a causal relationship with MTCT, then cessation of breastfeeding would need to be abrupt. However, there is evidence that poor maternal health, itself a risk factor for post-natal MTCT (Embree et al. 2000; Breastfeeding HIV International Transmission Study Group 2004), is associated with mixed feeding (Chisenga et al. 2005). Furthermore, abrupt cessation of breastfeeding at 4 months proved difficult for many Zambian women and was not associated with any increase in HIV-free survival at 24 months compared to the normal practice of longer breastfeeding and gradual weaning (Kuhn et al. 2008). In a small study where weaning was not recommended to be abrupt, there was no association between feeding practice and milk HIV RNA viral load (Phiri et al. 2006). Therefore, although weaning for most HIV-EU should be sooner than usual for HIV-unexposed infants in a population, there is not single age at which all HIV-infected women should wean and there appears to be no benefit to weaning abruptly.

Immune abnormalities

A number of studies have shown immune abnormalities among HIV-EU children. Appropriate controls for studies of lymphocyte populations are essential because of significant differences by race and gender (European Collaborative Study 2003b). HIV-EU European and American infants, most of whom had been exposed to ART, had several differences in proportions of blood T cell populations, the most consistent being fewer total CD4 and fewer naïve CD4 with variable CD8 and memory cells (Clerici et al. 2000; Embree et al. 2001; Nielsen et al. 2001). In a large cohort in Kenya, differences in CD8 cell numbers between HIV-EU and unexposed children disappeared after 1 year of age (Embree et al. 2001). In contrast, Italian HIV-EU children (aged about 6–7 years), none of whom had been exposed to ART, still had some abnormalities, for example, fewer naïve and more memory cells, than HIV-unexposed children, a cell profile in many ways intermediate between that of unexposed and HIV-infected children (Clerici et al. 2000). Fewer progenitor cells and lower thymus output, as estimated from the percentage of cells with T cell excision circles, appeared to account for the reduced number of naïve CD4 cells (Nielsen et al. 2001). Cell population percentages may also be altered because HIV exposure leads to increased immune system activation (Schramm et al. 2006) and subsequent T-cell apoptosis (Economides et al. 1998).

Table 1 summarises cytokine production abnormalities investigated among HIV-EU children. Overall, although there are many documented abnormalities in cytokine production by HIV-EU children, much variability is seen and it is not clear whether specific abnormalities could result in increased morbidity or mortality.

Table 1.   Cytokine abnormalities among HIV-exposed, uninfected children compared with unexposed children
CytokineAge at testingStimulusResultCountryReference
  1. HIV-EU, HIV-exposed uninfected, IL, interleukin, PHA, phytohaemagglutinin, IFNγ, interferon-gamma, PPD, purified protein derivative of tuberculosis, BCG, Bacille Calmette Guerin, ESAT-6, Escherichia coli antigen.

  2. Cytokine functions: IL-2: promotes T cell growth and differentiation; IL-4: stimulates lymphocytes and regulates humoral immunity; IL-7: lymphocyte development; IL-10: immunoregulation; IL-12: stimulates cell-mediated immunity; IFNγ: regulates immunity and inflammation.

IL-20–6 monthsPHADecreased T cell IL-2 productionUSA Rich et al. 1997
IL-71 month and 6–7 yearsNoneIncreased serum IL-7 at both agesItaly Clerici et al. 2000
IL-10Cord bloodPHAIncreased IL-10 productionSouth Africa Kuhn et al. 2001
IL-122–38 months Staphylococcus aureus Decreased IL-12 production by blood mononuclear cellsUSA Chougnet et al. 2000
IFNγ1 month and 6–7 yearsHIV-gp160 peptidesIFN γ produced in response to HIV peptides in 1 month HIV-EU onlyItaly Clerici et al. 2000
IFNγCord bloodPHANon-significant trend towards increased IFN γ productionSouth Africa Kuhn et al. 2001
IFNγ, IL-2, IL-4Cord bloodNoneDecreased levels of IFNγ but not IL-2 or IL-4Denmark Nielsen et al. 2001
IFNγ1 day and 6 weeksNone or PHA or PPD or ESAT-6 or BCGIncreased IFNγ in unstimulated whole blood cultures at both ages and in response to PHA at 1 day; no other significant differencesSouth Africa Van Rie et al. 2006

Functional immune assays, mostly related to tuberculosis, have also been investigated among HIV-EU children. HIV-EU Gambian infants were less likely than HIV-unexposed infants to develop a scar after BCG-vaccination (Ota et al. 1999) although normal scar formation was seen among South African infants (Van Rie et al. 2006). In Rwanda, HIV-EU children had normal BCG scar formation, but reduced skin test response to tuberculin (Msellati et al. 1991). It is unclear from these results whether HIV-EU children are more susceptible to tuberculosis, but, given that many are exposed to parental tuberculosis, this is a potential concern.

The causes of the immune abnormalities of HIV-EU children are unclear. Many of the immunological changes seem to result from exposure to HIV itself and thus specific responses to HIV are seen, even when infants remain uninfected (Kuhn et al. 2002). Exposure not only to virus but also to maternal placental responses to the virus has been suggested as a modifier of immune functions of HIV-EU infants (Clerici et al. 2000). Although direct evidence for this seems unavailable, it is known that HIV-infected, ART-treated women have greatly increased placental production of the inflammatory cytokines, tumour necrosis factor-α (TNFα) and interleukin-8, compared with uninfected women (Faye et al. 2007) and that raised placental TNFα in association with malaria can affect the foetus, as indicated by the association of TNFα with low birth-weight (Rogerson et al. 2003). Reduced placental antibody transfer, for example, of measles antibody (Farquhar et al. 2005a; Scott et al. 2005), could also compromise infant protection and thus affect infant immune development. Much of the data on immune function of HIV-EU children is from industrialised countries where these children are rarely breastfed. Breastfeeding promotes immune development (Hanson 2000) as well as protects against infections, which could alter immune status. Unlike placental antibody transfer, concentrations of immune factors in breast milk do not appear to be reduced in HIV-infected women (Shapiro et al. 2007) and may even be increased in association with the higher prevalence of subclinical mastitis (Filteau et al. 1999; Semba et al. 1999; Farquhar et al. 2005b; Kasonka et al. 2006). However, abnormalities are also seen among breastfed African HIV-EU children so lack of breastfeeding cannot be the only mechanism.

Exposure to other infections

In addition to exposure to HIV, HIV-EU children are at increased risk of exposure to other infections (Mofenson et al. 2005). HIV-EU children may be exposed to parental illness, for example, tuberculosis (Cotton et al. 2008) or P. jiroveci (Heresi et al. 1997) and to dietary pathogens if not breastfed.

A number of viruses are not normally pathogenic in healthy immunocompetent adults, for example, herpesviruses, including human cytomegalovirus (HCMV), but can be pathogenic in infants during acute primary infection and particularly in neonates with impaired immunity. These persistent viruses establish latent infections and can be reactivated in immunocompromised groups, including HIV-infected adults, because of lack of control by their immune systems; virus reactivation correlates with AIDS progression as well as poor prognosis in co-infected infants (Kovacs et al. 1999). The viruses can also reactivate during pregancy and HIV-infected mothers may be at particular risk of passing HCMV to their offspring. Importantly, recent studies show that HIV-EU infants can be infected with these opportunistic virus infections and this may affect their subsequent development (Bates et al. 2008, in press).

Exposure to antiretroviral drugs

Several studies have addressed whether exposure of infants to antiretroviral drugs given to either their mothers or the infants themselves have adverse effects. Most studies have been conducted in industrialised countries, so maternal exposure is in utero as few HIV-exposed infants are breastfed. Many antiretroviral drugs appear in pharmacologically active concentrations in breast milk (Colebunders et al. 2005; Shapiro et al. 2005), so African infants may have this additional exposure. In addition, many prevention of MTCT regimens include direct ART for the infant.

Perinatal exposure to ART appears responsible for some of the abnormalities in lymphocyte, neutrophil and red blood cell populations of HIV-EU children (Bunders et al. 2005a,b; Pacheco et al. 2006; Feiterna-Sperling et al. 2007). Some of these abnormalities persist for at least several months. The other major ART side effect of concern is mitochondrial abnormalities, usually associated with nucleoside analogue reverse transcriptase inhibitors and usually detected by screening for elevated plasma lactate levels (Noguera et al. 2004). The raised lactate is transient in most children, but in a few children, it persists up to 6 months and is associated with neurological symptoms, such as mild axial hypotonia or limb spasticity and poor head control, which later resolve (Noguera et al. 2004).

Although exposure to zidovudine for >7.5 weeks in utero was associated with lower birth weight, the effect faded with time since there was no effect of ART on post-natal growth among non-breastfed Thai infants (Briand et al. 2006). Similarly, in the European Collaborative Study, there were minimal growth differences between HIV-EU children exposed or not to maternal ART (European Collaborative Study 2005). In summary, although adverse effects of ART should be monitored, they are usual transient and do not seem serious enough to avoid giving ART, which has enormous benefits for both mother and child. Furthermore, health problems have been seen among HIV-EU children where ART has not been available so ART exposure is unlikely to be a major mechanism for health problems among these children.

Approaches to improving health and development of HIV-EU children

Figure 2 summarises potential mediators of the poorer health and survival of HIV-EU, compared with unexposed, children. Arguably it would make sense to concentrate efforts on ‘upstream’ factors to have the greatest benefit. Reducing the number of HIV-infected adults is clearly primary.

Figure 2.

 Summary of mechanisms which might contribute to poor health and survival of HIV-EU children.

A review written for WHO recommends a broad approach for psychosocial support of children and families affected by HIV (Richter 2006). There is potential for using the health services, particularly programmes for prevention of MTCT, as a means of accessing such families. It should be possible to include care and support of HIV-EU children within this context. This will, however, require considerable investment in health services since as yet only a minority of HIV-infected African women have access to even basic prevention of MTCT services (UNAIDS 2006).

Many of the interventions designed to support either the mother or any HIV-infected children will at the same time benefit HIV-EU exposed children. Medical support includes improving maternal health, including by providing ART, providing mothers with iron-folate or other micronutrient supplements and providing prophylactic antibiotics against opportunistic infections for infected mothers and for infants before HIV status is known. As there is evidence that orphans and vulnerable children can be protected within families and communities (Richter 2006; Parikh et al. 2007), programmes to support HIV-EU children should build on these successes by providing food, financial or community support for extended families caring for HIV-EU children. These children should also be involved with other children in group play, school or other activities, which promote cognitive and motor development.

Other factors adversely affecting health and development of HIV-EU are less amenable to intervention. Exposure to HIV is the main problem and cannot be avoided entirely; however, maternal health interventions, including ART, can reduce the intensity of exposure to virus.

In the end, the most challenging problem in the area remains how to promote the best infant feeding options for children whose mothers are HIV-infected. It seems that the feeding mode that is associated with lowest MTCT – early exclusive breastfeeding followed by weaning onto AFASS foods as soon as is feasible – is the mode which best promotes health of HIV-EU children.

Research gaps

The main research gap concerning how to promote health of HIV-EU children is for programmatic research. Programmatic research generally needs to be specific to particular locations and populations and thus requires local input into programme design. Programmes may not always be easily extrapolated to other groups, so programmatic research is often not seen as important to the scientific community. Programmes to improve health of HIV-EU African children are unlikely to provide commercial gain so are of little interest to industry. Nevertheless, good programmatic research is at least as difficult to design and conduct as good basic research and, with a need for large sample sizes, training, or improvements in infrastructure, often costs as much as basic research. Thus, funding agencies must value and prioritise programmatic research if we hope for health gains for HIV-EU children and their families.

We need to determine how best to support household food security for HIV-affected families in an ethical way, particularly in areas where household food security is poor for all families. We need interventions to improve children’s cognitive and motor development. There is increasing focus on including stimulation interventions in child health programmes (Engle et al. 2007). Given the importance of not segregating and thus possibly stigmatising, HIV-affected children (Richter 2006), wider spread of such stimulation interventions are likely to benefit all children together with HIV-EU children.

There is also a need for further basic research. Our own results, discussed above, suggest that breastfeeding capacity of HIV-infected women, even those motivated and supported to exclusively breastfeed, may not be equivalent to that of HIV-uninfected women (Chisenga et al. 2005). Research should address how maternal health interventions, both during pregnancy and post-partum, affect both breastfeeding practice and milk output. For example, multivitamin supplements with an emphasis on antioxidants given to HIV-infected Tanzanian women improved several aspects of maternal and infant health (Fawzi et al. 1998, 2004) including infant weight gain (Villamor et al. 2005) and motor development (McGrath et al. 2006) irrespective of infant HIV status. In spite of these promising results, such supplements do not appear to have been investigated or incorporated into programmes elsewhere.

Further research is required regarding locally available AFASS foods to give to children following about 6 months of exclusive breastfeeding. Complementary or replacement foods in much of Africa as well as in low income countries elsewhere, are frequently bulky with low energy and micronutrient density (Becquet et al. 2006; Lunney et al. 2008) and thus could not be considered as fulfilling the AFASS criterion of being nutritionally ‘safe’. Nutritious replacement foods may be expensive (Owino et al. 2008), which means they do not meet the AFASS criteria of being affordable or sustainable. As available foods and cultural traditions around infant feeding vary across populations, research on improving complementary feeding, unlike that on improving breastfeeding, must be location-specific. Thus, like programmatic research to improve support and care of HIV-affected families, it must be led by local researchers who understand their communities.

Further research should be directed towards how to reduce the exposure of infants of HIV-infected mothers to other infections, notably herpesviruses. Trials could determine whether provision during late pregnancy or early infancy of anti-viral agents for these viruses, in addition to antiretroviral drugs for HIV, can decrease infant infection and improve infant health and survival. Such studies could also investigate whether such interventions modify maternal placental inflammatory cytokine responses with consequent effects on the infant.

It will be important to follow up cohorts of HIV-EU children for longer periods to determine whether the common combination of low birth-weight and rapid post-natal weight gain, especially of fat, increases the risk of chronic diseases later in life. Promotion of cardioprotective diets and exercise might be beneficial in theory, but are likely hard to implement in many poor African communities.

Conclusion

As the HIV epidemic continues and ART becomes more widely available, HIV is becoming a concern for public health services as well as for clinical medicine services. In addition to ongoing medical support for HIV-infected children, there needs to be more holistic family and community support for the growing number of HIV-EU children. Health care staff need to be made aware and trained in managing public health interventions alongside clinical interventions for families affected by HIV.

Acknowledgements

The author’s cited work has been supported by the Wellcome Trust and by the Bill and Melinda Gates Foundation. I am grateful to Dr Ursula Gompels for contribution to the section on other infectious disease exposure.

Ancillary