By the mid 1990s Zambia’s prevalence had overtaken Malawi’s. According to a later estimate, adult infection peaked in 1994–5 at about 17 per cent. As in Blantyre, the growth of the epidemic in the late 1980s was especially rapid in Lusaka, where antenatal prevalence rose to a roughly stable 22–27 per cent at different clinics in 1990–3. The difference was that 50 per cent of Zambians were urban and that overall prevalence in Copperbelt province almost equalled that in the capital.13
Yet Zambia, too, was soon overtaken. Zimbabwe’s prevalence figures are especially difficult to interpret, with wide variations between those quoted by national and international authorities and even wider fluctuations at individual sentinel sites. The most reliable data are probably for antenatal clinic attenders in Harare. Prevalence among them was 10 per cent in 1989 and 18 per cent in 1991, both figures substantially less than in the main cities of Malawi and Zambia, but it grew further to a peak of 32 per cent in 1995 and then fluctuated around that level. Yet only 28 per cent of Zimbabwe’s people were urban.14 The distinctive feature of its experience during the 1990s was the high level of prevalence outside the main cities, often so high that the statistics must be treated with caution. Three kinds of areas were worst affected. One contained towns on main roads close to borders, where truck drivers might socialise for several days while negotiating their way across the frontier. Beitbridge, on the South African border, recorded 59 per cent HIV prevalence in 1996, while the figure at Mutare, near the frontier with Mozambique, reached 37 per cent in 1997.15 Second, the trucking routes contributed to high prevalence in provinces and districts through which they passed. Masvingo province, which registered a barely credible provincial figure of 49.4 per cent among pregnant women in 2000, was bisected by the road from Harare to South Africa, while Midlands province, with a reported 45.1 per cent prevalence in 2000, straddled the route from Harare to Bulawayo.16 Yet this devastating provincial infection that distinguished Zimbabwe was not confined to transport routes but existed even in remote rural areas. In 1993–4 overall adult prevalence was already 24 per cent in the Honde valley, a fairly isolated part of Manicaland. Shortly thereafter, 22 per cent of pregnant women tested HIV-positive even at Tsholotsho in arid northern Matabeleland.17 As will be seen later, both its excellent transport system and its high levels of oscillating migration between country and town made rural Zimbabwe especially vulnerable to infection.
Those characteristics operated even more powerfully in Botswana. From only 2 or 3 per cent in 1990 its national adult prevalence soared to 23 per cent in 1995 and either 28 per cent (according to the government) or 36 per cent (according to UNAIDS) in 2000, the latter figure being the highest in the world.18 As the epidemic spread south, its momentum seemed to accelerate, suggesting the possibility that rapid passage of the virus from person to person might be increasing its virulence, although there was no hard evidence of this. The acceleration in Botswana was noticed first not at the capital, Gaborone, but at Francistown, where the main road crossed into Zimbabwe and antenatal prevalence reached 24 per cent in 1992 and 34 per cent in 1993. Gaborone soon followed, as did the mining town of Selebi Phikwe; in 2000 these three towns registered antenatal prevalences of 44, 36, and 50 per cent respectively.19 Yet this initial urban predominance was reversed as the epidemic grew. By 1999 prevalence among pregnant women was 22 per cent even in the Kgalagadi desert area, while the highest reported prevalence among them at that time was 51 per cent in the northern district of Chobe. Overall, according to the government, ‘the 2002 survey reveals slightly higher rates in rural than in urban areas’. The annual incidence of new infections for the whole country at that time was estimated to be 6 per cent, roughly three-quarters of the level reached among young people at Rakai during the 1980s.20
An early attempt to explain the speed and scale of Botswana’s epidemic highlighted three factors: ‘the position of women in society, particularly their lack of power in negotiating sexual relationships; cultural attitudes to fertility; and social migration patterns’.21 Gender inequality fostered the epidemic throughout Central Africa. Commercial sex, driven mainly by female poverty and lack of opportunity, has been little studied in Botswana, but elsewhere it was important especially in initial urban epidemics, although probably less central than in Nairobi or Kigali. Women held only 8 per cent of Zimbabwe’s and 15 per cent of Zambia’s formal sector jobs in the early-mid 1990s.22 ‘Divorce, rural poverty and superior earnings were the principal reasons cited’ by sex workers in Harare in 1989; 70 per cent of them were divorced, probably with children to support, and nearly half came from drought-stricken southern Matabeleland. Six years later, 86 per cent of sex workers tested there had HIV, like 70 per cent of those working the main road between Zimbabwe and Zambia in 1987, 56 per cent in Blantyre in 1986, and 69 per cent in Ndola in 1997–8.23 Although willing to use condoms, only about half of those in Harare in 1989 and one-quarter of those in Blantyre and Ndola in the mid 1990s could overcome their clients’ opposition.24 Studies of young male factory workers in Harare during the 1990s showed both their fecklessness and their difficulty in avoiding risk where HIV was so widespread. Their annual incidence of new infections was 2 per cent, meaning that half were likely to contract HIV during a normal working lifespan. Similar levels of infection existed among long-distance drivers.25 A Malawian villager later recalled how passing tanker drivers infected local women:
The wives were spreading the virus to their husbands, the unmarried women were infecting the young men, the young men making money from smuggling were going into Lilongwe and having sex there. People were behaving very freely and they had no idea that anything bad could happen to them. . . . By 1996, 12 years after the trucks first started arriving, the death rate in the village peaked at four a week. . . . Our neighbours from other villages would not come to help people who were sick or help at a funeral because of fear of contracting the disease. . . . We became completely isolated.26
More commonly, however, infection passed from promiscuous men to their wives. In one small enquiry in Lusaka, lasting a year, 26 per cent of HIV-positive husbands infected their wives, while only 8 per cent of HIV-positive wives infected their husbands. ‘Men generally acquire infection first,’ a careful study in Manicaland reported, ‘frequently during spells of labour migration in towns or commercial areas, and then pass on the infection to their regular female partners based in rural areas.’ By 1998 twice as many women as men there were infected, including four times as many among people aged 17–24, owing to the disparity of age between sexual partners.27
Nevertheless, women too could be ‘movious’, as Central Africans described it. Most were not: even the highest self-reported accounts of sexual behaviour suggest that only about 25 per cent of women had non-marital sex. Yet of those attending antenatal clinics in two areas of Manicaland in 1993–4, 16 per cent of married women, 43 per cent of single women, and 50 per cent of formerly married women were infected.28 Among the many factors encouraging extra-marital sex, one of the most important was delayed marriage, due chiefly to education, labour migration, and the decline of polygyny. In Botswana in 2001, for example, the median age at first marriage or cohabitation was 28 for men and 23 for women. Consequently, in 1995 over 60 per cent of never-married women aged 20–24 there were mothers, while 41 per cent of boys and 15 per cent of girls aged 15–16 had sexual experience. In Lobatse and Francistown, with very high HIV prevalence, 47 per cent of men and 39 per cent of women aged 17–18 had a casual partner over a twelve-month period; 21 per cent and 16 per cent had at least two. Of teenage girls who bore children in the late 1980s, 40 per cent had them with men six or more years older than themselves. Young Tswana had adopted an experimental attitude towards sex – ‘marketing themselves’ as it was known – ‘so that you