Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

South Africa has one of the most extensive HIV epidemics in the world, and one where the burden of the epidemic is most conspicuously borne by young black women. In the interests of epidemiological pragmatism, its course has until fairly recently been mapped through women’s infections. One consequence of which has been to render men relatively invisible, both as HIV-related health-service users and as agents of sexual risk (Greig et al., 2008). Yet women’s subordinate position in a highly patriarchal society has critically shaped their HIV risk, just as the racial patterning of the epidemic has its roots in the political and economic subordination stemming from colonisation and the era of apartheid. This chapter is largely an account of influences on the lives of African women, who carry the overwhelming burden of infection.

Until 2002 when the Human Sciences Research Council undertook its first population-based survey of HIV prevalence, the subclinical (and clinical) course of the infection was followed in antenatal populations. Anonymous surveillance in antenatal clinics by the Department of Health has mapped an initial exponential rise in national prevalence of HIV among pregnant women from 0.7% in 1990, 16.0% in 1997 to 24.5% in 2000. The first decade of the millennium has seen the prevalence plateau and the most recent rate was 29% in 2007 (Department of Health, 2008). HIV does not equally affect all South Africans. There are marked racial differences, with national prevalence estimates showing that white and Indian men and women in the general population have very low prevalence (0.6% and 1.9%, respectively), while the highest is found in Africans (13.3%) (Shisana et al., 2005).

The sexually transmitted portion of the epidemic starts in the teenage years, when women begin to be infected. This starts approximately 5 years younger than infections in their male peers (Shisana et al., 2005). A national survey of youth aged 15–24 years found that 15.5% of women and 4.8% of men were infected (Pettifor et al., 2005). Women sustain this vulnerability throughout their lives. Among adults aged 15–49 years, nearly twice as many men as women have HIV (11.7% versus 20.2%) (Shisana et al., 2005). An inescapable conclusion drawn from these gendered patterns of infection is that the epidemic in women is disproportionately driven by the behaviour of a relatively small group of HIV-infected older men who have a large number of partners. Little wonder that studies describing factors associated with HIV in young women have consistently pointed to the importance of older partners (Jewkes et al., 2006a; Pettifor et al., 2005). Other identified risk factors in young women are having more partners (Jewkes et al., 2006a; Pettifor et al., 2005), more sex (Jewkes et al., 2006a) and getting sexually transmitted infections and using condoms inconsistently or not at all (Pettifor et al., 2005).

These factors reflect some well-recognised biological aspects to women’s vulnerability. HIV is more readily transmitted from men to women than women to men (Gray et al., 2001), with explanations related to the larger surface area of the vagina and frequency of tears during sexual intercourse. Older men are more risky for women because they are generally more sexually experienced and so are more likely to have STIs that can be transmitted to women. This is seen in their higher age-specific prevalence of HIV and HSV-2 (an established HIV co-factor) than the women’s male peers (Pettifor et al., 2005). STIs have been shown to increase vulnerability to HIV acquisition, particular herpes simplex type 2 virus which may convey as much as threefold increased risk (Freeman et al., 2006). In mathematical terms, having more partners increases the likelihood of having one who is HIV infected. While the biology is indisputable, it does not particularly help answer the important questions about the social and cultural origins of and social processes that sustain patterns of sexual practices that place women at risk of HIV. These questions relate to a level of causation that is deeply rooted in the social fabric of life in South Africa, the understanding of which is absolutely critical for the development of effective approaches to reduce sexual risk taking. In this chapter, we reflect on major historical processes that have influenced the shape of present-day sexuality and discuss some of the impacts of apartheid and urbanisation. Recognising women’s sexual practices as both influenced by and influencing a broader set of ideas about womanhood (or femininity) which are shaped by sets of power relations that critically pertain between women and men, as well as between women and others in their social environment, we then discuss sexuality in the context of constructions of femininity, gendered power relations and gender-based violence. We conclude with thoughts on the implications of this for reducing women’s vulnerability to HIV.

2 Legacy of Apartheid, Patriarchy and Urbanisation: Historical Perspectives

The dominant feature of the socio-political landscape of South Africa in the last century was the growing institutionalisation of racial discrimination, which culminated in the policies of apartheid. So perhaps it is inevitable that a narrative that seeks to locate women and the modern-day epidemic of HIV within a broader social context should find roots within the major social influences of apartheid and urbanisation. These impacted very directly on the position of women in society. Pre-colonial South Africa was characterised by the subordination of women to men, but the convergence of this society with emerging capitalism and the legislative framework of institutionalised racial discrimination led to transformation in gender relations with massive accentuation of gender inequalities, most visible in new black urban areas where women were mostly totally dependent on men (Walker, 1990). These processes at a fundamental level influenced the position of sex in society: how it was viewed, its relationship to marriage and family life and women’s sexual power.

Discourse and practices related to sex and sexuality in southern Africa are characterised by an interplay of two, very contrasting, sets of ideas. One views sex through a lens of missionary prudery, where it is seen as requiring control and repression (Epprecht, 2004). A contrasting set of ideas, rooted in southern African cultural traditions, is characterised by a degree of openness and frankness about sex. Sexualised games played by children and ribaldry between adults and children have been a long-standing feature of normal childhood in many parts of the country. Authors, writing about quite different historical moments and contexts, have described ‘sex play’, which sometimes includes penetration, as very common among girls and boys from a young age (about 6 or 7 until early teenage years) (Longmore, 1959; Mager, 1999; Ntlabati et al., 2001). Indeed, sexualised play between adults and children has been described as an established feature of rural black South African childhood (Jewkes et al., 2005). Historically, dating and sex have been seen as key aspects of African teenage socialisation and have been accommodated socially within youth peer group activities and structurally within houses that were available for them (Mager, 1999). There was a prohibition, however, on full penetrative sex, which was driven by a need to prevent pregnancy. This was historically regulated by peers and those found practicing it were punished by them (Mager, 1999; Mayer and Mayer, 1970). The growth of urbanisation in the second half of the twentieth century, and spread of Christianity, however, brought with it both restrictions on the space available for open discussion of sex between generations and sexes, and a rejection of limited sex, which was seen as old fashioned (Mayer, 1961). Thus, in a somewhat contradictory way, sexual openness which had previously provided the opportunity to regulate sexual activity of young people was constrained, at the same time that forces of modernity were pushing young people towards patterns of sexual activity that were much more risky for them in terms of both pregnancy and sexually transmitted infections.

Historically, full intercourse was intended to be confined to marriage so as to prevent pre-marital pregnancy. Marriage after loss of virginity resulted in the bride wealth (lobola) being reduced in value, but historically there has been no expectation that sex should be confined to marriage or that after marriage men should stick to one partner. The last 50 years, however, have seen pre-marital pregnancy become the norm for Africans, and half of women become pregnant before their 21st birthday (Department of Health, 2007). The present-day median age at marriage is 28.5 years for women (Department of Health, 2007), but many women never marry or are only intermittently partnered in their lives. A total of 42.4% of households in South Africa are female headed, which reflects the high proportion of adult women who at any stage are living, very often with their children, without cohabiting men (Department of Health, 2007).

The late age at marriage and high prevalence of children being raised without fathers are both part of the legacy of apartheid (Wilson, 2006). Throughout the twentieth century, migrant labour was a key foundation upon which the wealth of the apartheid state was created. Legislation regulated residence in urban areas through the influx control (pass) laws, and these prevented rural wives and families living with men in ‘white’ areas without work or permits. Fathers became increasingly absent figures from home, and thus played ever-diminishing roles in the lives of their children. Indeed, many children were raised by grandparents and other relatives in rural areas, as low wages coupled with demands of women, and often new families in urban areas, meant that remittances home were infrequent or ceased, and many mothers themselves went to seek employment in urban areas (Delius and Glaser, 2002). Unmarried men found that low wages meant they had to work longer to raise bride wealth, and historians argue that from the 1950s marriage became increasingly unaffordable for an average working-class family and increasingly cohabitation and child-rearing in other family contexts were socially tolerated (Delius and Glaser, 2002; Hunter, 2006). A further consequence of migration for women was the tacit recognition that both partners would need to satisfy their sexual needs outside of a main relationship during periods of separation, which promoted practices of having multiple concurrent partners (discussed in Chapter 2). The extent to which women, as well as men, took advantage of this can be seen in the finding of research that among sero-discordant migrant mine workers from Hlabisa, in KwaZulu Natal, one-third of the infections over the period of observation were found to occur in the wives of the migrant workers (Lurie et al., 2003).

In the cities, life was tough. Work was scarce, poorly paid and often dangerous, and workers faced violence and humiliation at work, in daily encounters with apartheid laws and the enforcing police. In this context, many of the migrant men from rural areas, together with the increasingly established urban population, adopted more exaggerated constructions of masculinity, predicted on the control of women, conspicuous displays of strength, the use of violence and risk taking. Such ideas of masculinity were forged in the harsh environment of mine hostels, urban gangs and prisons (Delius and Glaser, 2002). Whilst there were at any one time competing constructions of masculinity, and extremes were seen in the harshest conditions among men in gangs and prison, the core elements of power and control of women, a willingness to use of violence, and elements of sexual and other risk taking were fairly ubiquitous (Morrell, 2001).

Whilst the impact of apartheid on families forced some women into patterns of living that entailed considerable separation from men or male partners, poverty drove other African women into relationships of dependence. During the apartheid years, there were numerous barriers to black women joining the labour market, not least Bantu educationFootnote 1and restrictions on access to tertiary-level study. Just after the advent of democracy, in 1995, 39% of workers were women (Statistics SA 1996), a statistic that changed little over the next decade, and generally women were also less well paid (Casale and Posel, 2005). This economic reality, coupled with the frequent absence of children’s fathers, has placed many women in an economically vulnerable situation, which could ultimately be made tolerable through attachments to available men (Hunter, 2002). Thus, poverty has driven many women into relationships of financial dependence, in which they have very little power and are open to exploitation and abuse, whilst at the same time having very limited ability to exert the ultimate sanction of leaving.

3 Women, Femininities and Sexuality

The emerging body of literature on sexuality in South Africa, a product of research over the last 15 years, has overwhelmingly tended to discuss gender in terms of violent and controlling masculinities and passive, victimised femininities. Indeed, in the health literature, women’s sexuality is predominantly discussed in terms of what women do (or what men do to women) and much less attention is given to the broader web of power, agency and meaning within which women’s sexual actions and experiences are embedded. Whilst there has been a very appropriate recent increase in awareness of the impact of gender inequalities on women’s health, women’s sexuality cannot be properly understood without reflection on alternative narratives of female agency and power, albeit constrained by the context of overwhelming male power, that have struggled to emerge, but form an important part of the landscape of sexuality in South Africa. Understanding these is also crucial for understanding women’s risk of HIV.

For teenage girls and older women, boyfriends are a means through which esteem is evaluated, and provide an irrefutable testament of beauty and desirability, which are generally seen as positioned centrally within dominant ideals of femininity. Coercion notwithstanding, both teenage girls and boys are generally active agents in their sexual relationships (O’Sullivan et al., 2006), and research on teenage sexuality shows that by the age of 17, half of all teenagers are sexually active (Manzini, 2001). Relationships are also mechanisms through which teenage girls come to experiment with and explore their power as women (Nduna and Jama, 2000). In a context where most African teenage girls grow up with very limited access to material resources, yet are constantly bombarded with messages that promote conspicuous consumption, beauty and desirability are deployed to leverage resources through their relationships, whether these be access to opportunities to party and have fun, or material goods such as a cell phone, airtime, cosmetics or, most often, cash (Dunkle et al., 2004a). These may then be deployed in competitions for status where women seek to gain the respect and admiration of other women.

There is a growing literature on ‘sugar daddies’ and transactional sex, which differentiates relationships where there is a strong expectation of material gain from activities of prostitution, where there is an agreed fee (Hunter, 2002). In South Africa where the norm is of gift giving by men to women who are romantic partners, and to some extent vice versa, there can be difficulties distinguishing predominantly transactional sexual relationships from those established for reasons of romance and desire. Affirmative motivations for relationships are generally found on a spectrum between love and pure transaction that spans feelings of affection, affirmation, reflected status, possibility of future financial security and obligation. Whilst researchers often define transactional relationships as ones that would not exist if it were not for the material exchange, they may nonetheless be driven by several of these motivations at one time, and these may shift (Dunkle et al., 2004a). A sense of the commonness with which African women have transactional sexual relationships was given by the finding that 21% of women in interviews with a Soweto antenatal population disclosed having had transactional sex with a non-primary partner (Dunkle et al., 2004a). Research with impoverished rural men shows that they not infrequently have transactional relationships based on their receipt of gifts from women too (Dunkle et al., 2007).

Whilst women may engage in these relationships in an active process of securing material benefits and exerting power over their lives in contexts of poverty (Hunter, 2002), the realities of the power dynamics are such that they have severely limited capacity for influencing the terms on which sex takes place once the relationship is established (Dunkle et al., 2007). Economic dependency that is definitionally entailed in transactional relationships provides one set of constraints on women’s power, but still others stem from the nature of the men who engage in them. Given high levels of youth unemployment, men who are positioned to have transactional sexual relationships are usually older and more powerful than their women partners, or otherwise engaged in crime and violence. South African society is structured by dominant age and gender hierarchies, which doubly disadvantage women in age-disparate relationships. Transactional sex has been shown in men to be associated with a range of other risky sexual behaviours, which increase the likelihood of the men who engage in it being infected with HIV (Dunkle et al., 2007), so it is not surprising that women who have had transactional sex are more likely to have HIV than other women (Dunkle et al., 2004b).

Relationships with older men also very commonly co-exist, on the woman’s part with more equitable relationships with peers and on the man’s part often with a wife or main partner, and so feed into complex sexual networks through these multiple points of concurrency. This enhances the likelihood of spread of HIV if it is present in the sexual network (Epstein, 2007). Partner concurrency is common for men and women. Whilst endeavouring to have a main partner, who is publicly known as their boyfriend, women may also have secret partners (termed khwapheni in Xhosa or nyatsi in Sotho) with whom they illicitly meet for sex, sometimes over many years (Selikow et al., 2002). This is very common when men and women have had a child together. They may continue to have this sort of relationship long after they ceased to be each other’s main partner (if indeed they ever were), particularly if some form of financial support is given (Dunkle et al., 2004a). There is often a transactional element to relationships with khwapheni, if not always, but the element of concurrency is always present on the part of one or both partners. For women who are unhappily married, or have boyfriends who are violent, or otherwise do not meet their emotional needs, khwapheni may provide the source of love and affirmation they need in their lives. Counting sexual partners of South African women in surveys is often inaccurate as these relationships are often not reported unless specially asked about; the same applies to once-off sexual encounters. Qualitative research, which followed methodologies where women are befriended and interviewed over long periods of time, has shown us that khwapheni and once-off relationships are much more common than surveys usually report. Amongst pregnant women in Soweto, 40% reported having had a khwapheni in the last year (Dunkle et al., 2004a).

The complex patterns of sexual partnerships reflect the very great importance attached to them in women’s lives. Having a male partner is a central feature of South African femininity and seen as essential for validating self-esteem and peer-group status. Having more than one partner is often referred to as ‘walking on two legs’ (Jama and Jewkes, 2002), with the implication being that this is a stronger position as it reduces the chances of being without a partner. Having a main partner is particularly important as it positions women to enjoy the economic and social security of cohabitation, and possibly marriage. There is movement between the partner categories, and research shows that women who are khwapheni may be promoted to main partners, and so there is considerable competition between partners. This may occasionally break into the open in the form of fights, but more commonly, competition is covert, and may take the form of trying to prove that sex is better with them. In this context, insisting on condom use is a strategy that few women would risk. This partly explains the well-documented resistance of women, as well as men, to condom use (Wood, 2003).

Ethnographic researchers, such as Janet Wojcicki and Cathy Campbell, have described the lives of women engaged in sex work in mining areas and central Johannesburg as well as transactional sex in small towns (Campbell, 2000; Wojcicki, 2002; Wojcicki and Malala, 2001). These accounts, on the one hand, reflect the desperate vulnerability of many South African women with their childhoods of abuse and neglect, and their moves from these into contexts of survival through selling sex. On the other hand, these accounts also powerfully reflect the way in which, in a context of severe hardship and very limited options, selling sex may be a decision women make and an option for a lifestyle with features that are less awful than a range of possible alternatives (Campbell, 2000). Whilst describing selling sex for survival, these accounts are also infused with images of female solidarity and the enjoyment of time spent drinking and hanging out with women friends. Although there are powerful images among South African women of female chastity, obedience and abstinence from alcohol, which are ideals of femininity that are heavily infused with Christian morality, these compete with more traditional models of femininity which accord women more social, sexual and economic freedom (Epprecht, 1993; Gaitskell, 1982; Marks, 2002). Indeed norms of alcohol consumption among women who drink are quite high, and drinking and socialising often provide a context in which risky sexual encounters occur, and often casual sex, motivated by both reciprocity for drinks bought and sexual desire (Crush and Ambler, 1992).

Whilst there is considerable heterogeneity among South African heterosexual femininities and their construction in their relation to male power (as there is among masculinities, see Chapter 4), they are substantially positioned on a spectrum ranging from marked subordination to femininities that entail areas of resistance to male control of the forms described above. The epidemiological evidence does not indicate that the common ways in which women assert sexual agency that have been discussed here translate into agency that is useful in protection from HIV. Indeed, having multiple partners, transactional sex or older male partners and drinking alcohol are generally found to be highly risky (Dunkle et al., 2004b, Jewkes et al., 2006a, Pettifor et al., 2005). It seems likely that the reason for this is that these acts of sexual agency do not in themselves challenge the dominant structure of gendered power relations. Notable is how infrequent, at any level of South African society, it is for femininities to be constructed in relation to male power in a way that is largely premised on gender equity and an absence of control of either partner. This is explained by the absolutely central role of control of women in constructions of masculinity (Delius and Glaser, 2002; Morrell, 2001; Wood and Jewkes, 2001).

4 Men and Control: Gender Inequities and Gender-Based Violence

Delius and Glaser (2002), in their work on historical perspectives on sexuality, argue that the emphasis on masculinity in power and control over women has been part of processes of socialisation of men throughout the period for which historical evidence exists. This intensified and became increasingly violent in the second half of the twentieth century. They argue that sexual violence directed against women who were not partners became particularly prominent in urban areas from the 1930s with the growth of gang culture, and there were multiple accounts of gangsters perceiving that women who lived in their territory were their sexual possessions (Delius and Glaser, 2002). The accounts of ethnographers in rural areas point to very similar assumptions of sexual entitlement among rural men (Delius and Glaser, 2002; Schapera, 1933).

The growing body of research on rape in South Africa suggests that the core elements of these ideas remain highly prevalent today. South Africa has the highest rate of rape reported to the police of any country in Interpol (Interpol, 1996). Research with men has found that a very high proportion disclose having raped. In one study of young men from the rural Eastern Cape, 21% disclosed rape of a girlfriend or another woman, and the most commonly reported act was of gang rape (Jewkes et al., 2006b). Qualitative researchers have observed that young women often had limited opportunities for determining the timing and circumstances of sex and particularly first sexual encounters are quite commonly forced or pressurised (Jewkes et al., 2001; Wood et al., 1998). Within sexual relationships, women generally perceive that they have very limited scope for declining requests for sex. Sexual coercion has been reported as an experience of about one in four pregnant women in Soweto (Dunkle et al., 2004b), and by approximately two-thirds of sexually active teenagers in research in Cape Town (Jewkes et al., 2001).

Men often use physical violence to secure sexual submission and assert control over women. In dating and marital relationships, very high levels of physical violence have been reported by women in studies, often disclosed by 40–50% of women (Dunkle et al., 2004b; Jewkes et al., 2001, 2006a). Men also engage in a range of strategies to assert a position of dominance and control within relationships, which may not entail the use of physical violence or not require it to be used very often. These include threats, verbal abuse, asserting a right to be told where she is at any time, controlling access to friends and family, demanding attention or sex when requested and expecting her to be at home whenever he visits. Controlling behaviours often also include accusations of infidelity and monitoring of all contact with other men, expectations of not being challenged on one’s own infidelity and refusing to use condoms.

One of the most commonly mentioned barriers to condom use by couples is that they are not needed because they trust each other, and research has shown that this is commonly asserted by partners who are well aware that their relationship is or has recently been non-monogamous. Whilst, on the one hand, discourses of trust appear to reflect an aspiration of how relationships should be, research suggests that men who argue that condoms are not needed because of trust have been shown to be generally much more controlling of their partners (Shai et al., in press). Trust appears to be deployed as a manipulative strategy by which to further entrench the sexual control of women (Shai et al., in press). In similar ways, research has found that teenage women often report their partners forbidding them from using contraception and begging them to get pregnant to ‘prove love’ (Wood and Jewkes, 2006), despite the norm being that men take little financial or other responsibility for children fathered in such circumstances. Indeed, seeking a teenage pregnancy on a man’s part appears as one of the group of related practices that are indicative of a very gender-hierarchical ideal of masculinity, including having multiple partners, with payment for sex, controlling the behaviour towards female partners, with violence if necessary, and alcohol abuse (Hunter, 2005).

There is a growing body of epidemiological evidence that violent and controlling behaviours of men are associated with HIV risk in women. Research by Dunkle et al. (2004b) with women who were in antenatal care in Soweto aged 16–44 years showed the likelihood of having HIV to be elevated in women who had relationships characterised by greater gender power inequity, and who experienced physical or sexual intimate partner violence. They were also more likely to have HIV if they had used drugs, drank heavily, had more lifetime partners and had engaged in transactional sex. An analysis undertaken on a data set generated from interviews with much younger women (mean age 19, range 15–26 years) found that having more partners, more frequent sex or an older or more educated main partner was associated with greater HIV risk, and intimate partner violence was more commonly experienced by women with each of these risk factors (Jewkes et al., 2006a).

Violent and controlling practices impact on women’s risk of HIV in multiple ways. On the one hand, women who are raped are vulnerable to acquiring HIV in the course of the rape. Whilst this is of huge concern to women, and compounds the violation of the rape, desk-based modelling suggests that it probably does not result in many HIV infections each year (probably less than 0.05% of new infections) (Christofides et al., 2006). More important seems to be the general climate of powerlessness that is engendered by violent and controlling male behaviour. Indeed, there is considerable evidence that women who experience this have more frequent sex, and thus more opportunities for infection, and are less likely to use condoms (Dunkle et al., 2004b; Jewkes et al., 2006a; Pettifor et al., 2005; Shai et al., in press). Experience of rape, including abuse in childhood, and physical violence has a well-documented, long-term impact on substance abuse and sexual risk taking (Dunkle et al., 2004c; Jewkes et al., 2006a) compounded by the increased risk of re-victimisation associated with certain kinds of sexual risk taking (Dunkle et al., 2004c; Wojcicki, 2002). Post-traumatic stress disorder (PTSD) has been hypothesised as providing part of the explanation for this, as women reach for alcohol and other substances as a way of coping with the debilitating, and if untreated, long-standing symptoms (Wang and Rowley, 2008).

Men who are violent and controlling are inherently more risky for women because they are more likely to be infected with STIs, including HIV. Recent research from a random sample of men in the community (aged 18–49 years) found perpetration of intimate partner violence significantly associated with HIV (Jewkes et al., 2008a). In understanding the relationship between violent and controlling behaviours of male partners and women’s risk of HIV, it is critical to see these as part of a construction of masculinity, which also entails other risky sexual practices, including having multiple partners, engaging in transactional sex, coercing non-partners into sex, heavy drinking and drug use (discussed further in Chapter 4) (Dunkle et al., 2006, 2007; Jewkes et al., 2006b). Indeed, it was particularly interesting that the HIV-prevention behavioural intervention Stepping Stones, which sought to prevent HIV by building more gender equitable relationships, not only reduced men’s sexual risk taking tangibly (as shown by a reduction in new herpes infections over 2 years follow-up in a randomised controlled trial) but also reduced perpetration of intimate partner violence, sustained to 2 years after the intervention, and impacted on alcohol consumption, drug use and transactional sex (Jewkes et al., 2008b).

5 Prevention of HIV in Women, Challenges and Future Directions

African women in South Africa are in the forefront of the HIV epidemic, carrying the largest burden of illness, caring and other social responses. Preventing the continued spread of the virus amongst women is a very high priority. In recent years, there has been an overwhelming emphasis on the development of gender-sensitive technologies, with the hope that if successfully designed and deployed these will enable women to be empowered to protect themselves from HIV. Yet analysis of the broader social context of sexuality and HIV risk in South Africa suggests that biomedical interventions, which can be used by women without male cooperation, are unlikely to make much difference because they will not address the underlying constructions of masculinities and femininities that make such interventions necessary in the first place. Similarly, it is quite unclear what contribution interventions, which have been shown to reduce HIV infections in men, such as male circumcision, will have on HIV prevalence among women in the longer term. Broader transformative programmes that link promoting gender equality, economic and social empowerment and preventing sexual risk taking are also needed.

The apartheid legacy of poverty, poor educational attainment, lack of opportunities and high youth unemployment critically shape women’s perceptions of their lives, relationships and sexuality. These influences undoubtedly play an important role in explaining, for example, the very marked racial differences in HIV prevalence that are found in South Africa (Shisana et al., 2005). Structural change that strengthens economic opportunities for women and men is of vital importance. The IMAGE study evaluated an intervention, which included microfinance, a 10-hour attitude- and behaviour-change intervention focusing on gender and HIV that was delivered to groups of women at their two-weekly loan meetings and community action around violence against women (e.g. protest marches). The randomised controlled trial showed that programme participants in the intervention arm experienced 55% less intimate partner violence, although they were not shown to change their condom usage. This study is an important example of the potential impact of structural interventions on women’s lives and HIV risks (Pronyk et al., 2006).

Analysis of the broader context of women’s HIV risk suggests that transforming gender roles in a manner that increases gender equity and promotes health is vitally important. While the scale of change needed to fundamentally transform gender norms may seem overwhelming, there are signs of hope and promising strategies on the horizon. Research in South Africa shows that gender relations and the degree to which men and women negotiate aspects of sexual encounters can be made more equitable, have changed over the last decade and – importantly – can be changed by interventions (Jewkes et al., 2008b; Reid and Walker, 2005). The major contribution to sustainable, long-term HIV prevention could be made if it were possible to successfully promote such changes. Changing men is clearly critical for HIV prevention in women. Work with women is essential to ensure that as ideas about gender change women support these changes and are empowered to reduce their HIV risk.