Читать книгу The African AIDS Epidemic - John Iliffe - Страница 12

4

The Drive to the East

Eastern Africa was probably the first region to which HIV was carried from its western equatorial origin, along several different routes that cannot now be traced in detail. The virus entered a region divided historically into two contrasting natural and social environments: the well-watered, densely peopled kingdoms around Lake Victoria and on the Ethiopian plateau, and the less centralised societies in the drier savanna country where population clustered only on highland outcrops, in colonial cities along transport routes, and on the Indian Ocean coast. This framework gave HIV/Aids in eastern Africa its distinctive contrast between explosive epidemics in the Lake Victoria basin and the capital cities, on the one hand, and slow penetration into the remainder of the region, on the other. Varying relationships between cities and countryside were especially important in the process, as were the mobile groups linking them together and the factors – widespread labour migration, male predominance in urban populations, low status of women, lack of circumcision, and prevalence of sexually transmitted diseases – that bred higher levels of infection than in western equatorial Africa.

The virus first entered the Lake Victoria basin bordering the DR Congo. Patients from Rwanda and Burundi were seen alongside Congolese in European hospitals during the late 1970s and early 1980s. They not only led expatriate researchers to visit Kigali as well as Kinshasa in 1983 but encouraged observers of the epidemic to believe that Rwanda, Burundi, and perhaps even Uganda had been simultaneous or even earlier places of origin alongside western equatorial Africa. The location of the chimpanzee host makes this unlikely, however, as does the distribution of HIV-1 subtypes, for there is no indication in the Lake Victoria basin of the diversity of strains found in the DR Congo. Until well into the epidemic, the A and D subtypes dominated the region.¹

In Rwanda the first probable case recorded was a mother who displayed characteristic opportunistic infections in 1977 and subsequently tested positive for HIV along with her husband and three children.² A retrospective study found that by 1982 some 12 per cent of blood donors in Kigali were infected. The team visiting the hospital there a year later identified symptoms of Aids in 26 patients.³ The virus had apparently established itself during the 1970s and reached epidemic proportions by the early 1980s. The evidence from Burundi is even stronger, for 658 blood specimens taken during a study of haemorrhagic fever there in 1980–1 later revealed an HIV prevalence of 4.4 per cent, reaching 7.6 per cent in Bujumbura and 2.8 per cent in the countryside, at a time when Kinshasa’s antenatal prevalence was only 3 per cent. During 1983 cryptococcal meningitis, Kaposi’s sarcoma, tuberculosis, and other opportunistic infections became increasingly common in Bujumbura and doctors suspected Aids, which was confirmed serologically in 1984.⁴

Map 2 Eastern Africa

Bujumbura’s epidemic grew remarkably fast during the early 1980s. By 1986 some 16.3 per cent of women tested at antenatal clinics were infected. Thereafter growth slowed temporarily, rising only to an urban prevalence of 18.3 per cent in 1992, which nevertheless implied a high incidence of new cases.⁵ One reason for the epidemic’s virulence may have been its close association with tuberculosis, long prevalent in Burundi. In 1986, 55 per cent of tuberculosis cases treated in Bujumbura were HIV-positive, while tuberculosis cases in Burundi as a whole increased between 1985 and 1991 by 140 per cent. The epidemic’s most striking feature, however, was its urban concentration. While urban antenatal prevalence in 1992 was 18.3 per cent, it was only 5.2 per cent in semi-urban and 1.9 per cent in rural areas.⁶ Rwanda’s first rough sample survey of people of all ages in 1986 showed a similar contrast between 17.8 per cent prevalence in towns and only 1.3 per cent in the countryside. The highest rates among pregnant women were in Kigali, where they rose even more quickly than in Bujumbura, reaching 33 per cent in 1993.⁷

The rapid infection of Kigali and Bujumbura took place in countries where sexual behaviour among the overwhelmingly Christian general population was remarkably strict. In a survey conducted during the late 1980s, only 10 per cent of men and 3 per cent of women aged 15–19 in Burundi reported sexual intercourse during the last twelve months, compared with 51 per cent and 30 per cent respectively in the Central African Republic.⁸ The result was a different epidemic pattern, dominated not by widespread partner exchange but by commercial sex. Sex workers had long been Africa’s urban witches, blamed for all manner of social ills, so that there is a danger of stereotyping their role in the epidemic. Yet everywhere in eastern Africa, except Uganda, they were the first focus of infection. In 1984 a study of 33 sex workers in Butare, Rwanda’s second town and home to a military base and university, found 29 infected with HIV, along with 28 per cent of their clients, who frequented a median number of 31 sex workers a year. Of 300 Aids patients in Bujumbura in 1987–9, 106 of the 184 men had frequented sex workers and 21 of the 116 women had themselves been sex workers.⁹ Both countries were overwhelmingly monogamous, with exceptional numbers of unmarried women. In the early 1990s Kigali had 50 per cent more men than women aged 20–39. Fifteen years earlier the city also had an estimated minimum of 2,000 femmes libres, many of them uprooted by the destruction of Tutsi power since the revolution of 1961.¹⁰ On average, men in Kigali made their sexual debut at 18 but married at 24–28; in the meantime, since other young women in this ‘austere Catholic town’ were carefully protected, they frequented sex workers and often continued to do so after marriage. Circumcision was rare, condoms despised, sexually transmitted diseases widespread, sexual coercion common, and women depended overwhelmingly on a male partner for income.¹¹ Although epidemics in both countries initially focused around sex workers, therefore, their clients quickly spread the disease to their regular partners. In the late 1980s, 80 per cent of infected women and 76 per cent of infected men in Kigali had an infected partner.¹²

On the eve of the genocide of 1994, antenatal prevalence in the Rwandan countryside – ‘in the hills’, as they said in Kigali – remained less than 5 per cent. In rural Burundi it was even lower.¹³ The contrast with the relatively equal urban–rural prevalence that will be seen in Central Africa is difficult to explain in small countries with excellent transport systems, dense rural populations, and large income differentials between town and country. The towns did spread infection to their rural environs. A study in the Butare region in 1989–91 showed no association between HIV prevalence among rural women and the frequency with which they visited the town, but a significant association if their regular partner visited it daily.¹⁴ Yet these were small towns. The largest, Kigali, had only 220,000 inhabitants in 1986, only some 3 per cent of Rwanda’s population. They had no industry to attract the long-staying migrant workers who were probably most responsible for spreading infection to Central African villages. Whatever the reason, Aids in Rwanda and Burundi began and remained until the mid 1990s essentially an urban disease.

The contrast elsewhere in the Lake Victoria region was remarkable. In the lakeshore districts of Masaka and Rakai in south-western Uganda and the Kagera region of north-western Tanzania, Africa experienced its first rural-based Aids epidemic, a product of a prosperous peasant society at a moment of profound crisis. In East Africa during the 1970s the post-independence order was beginning to unravel. General Amin seized power in Uganda in 1971, precipitating eight years of violence and a magendo economy of illegality and self-help until the Tanzanian invasion overthrew him in 1979. Tanzania, although politically more stable, suffered severe economic decline as a result of the socialist strategy adopted in 1967, a decline accentuating Kagera’s long-standing problems of isolation, land scarcity, and agricultural decay. In Kenya, too, the prosperous era of Jomo Kenyatta gave way from 1978 to growing stringency and corruption under Daniel arap Moi.

HIV penetrated first into the borderland between Uganda and Tanzania west of Lake Victoria. Some have believed that the virus had been present in Uganda since the late 1950s or 1960s, pointing especially to occasional cases of the aggressive form of Kaposi’s sarcoma later found in some Aids patients. This is possible, but aggressive Kaposi’s sarcoma was a consequence of immune suppression rather than necessarily of HIV; nobody at the time noticed any change in the epidemiology of the disease, as they did in the early 1980s, and no stored blood from the region prior to the late 1970s has shown HIV antibodies.¹⁵ Without stronger evidence it seems more in accord with the continental pattern of the epidemic to think that Aids first appeared on the Uganda–Tanzania border in the late 1970s and HIV a few years earlier.

Its arrival cannot now be identified exactly. According to Uganda’s chief epidemiologist, symptoms later characteristic of Aids were first reported late in 1982,

when several businessmen died at Kasensero, an isolated small fishing village on Lake Victoria. This small town was also known for smuggling and illicit trade, and when these deaths occurred fellow traders shrugged it off as witchcraft. Others thought it was natural justice against those who had cheated. The only common characteristic the victims had was that they were all young and sexually active and stayed away from home for several days chasing wealth and presumably using it generously for their recreation and merriment.¹⁶

Across the border in Tanzania the first three Aids cases in Kagera region entered hospital in 1983. Retrospectively, however, medical workers believed that they had seen earlier cases. Kitovu Mission Hospital in Masaka district of Uganda was later said to have recorded 84 during 1982 alone. An African doctor in Rakai district believed that his uncle had died of the disease in 1980. The leading expatriate specialist later thought he had seen the corpses of Aids victims in Kampala in 1979 or 1980. Taken as a whole, the evidence suggests that HIV entered the region during the 1970s and became epidemic in the early 1980s.¹⁷

Local people called the new disease ‘Slim’ because wasting was commonly its most visible symptom. ‘In the first six months,’ Dr Anthony Lwegaba reported from Rakai in 1984,

the patient experiences general malaise, and on-and-off ‘fevers’. For which he may be treated ‘self’ or otherwise with Aspirin, chloroquine and chloramphenicol etc. In due course, the patient develops gradual loss of appetite.

II. In the next six months, diarrhoea appears on-and-off. There is gradual weight loss and the patient is pale. Most patients at this point in time will rely on traditional healers, as the disease to many is attributed to witchcraft.

III. After one year, the patient develops a skin disease . . . which is very itchy. Apparently it is all over the body. The skin becomes ugly with hyperpigmented scars. There may be a cough usually dry but other times productive.

IV. Earlier on after a year, the patient may be so weak that even when taken to hospital (not much can be done due to late reporting), goes into chronicity and death.¹⁸

Like the local people, Lwegaba blamed Slim on the young fishermen and smugglers who had flocked to the lakeshore to exploit the Nile perch fisheries and the magendo economy. ‘Since perch-fishing began,’ an investigator noted,

temporary fishing camps of grass huts and sheds have grown up seasonally on the lakeshore, with predominantly male populations. Male labour relies, for food, drink and sexual services, on cafés, teashops, and bars, largely run by women. Each camp is associated with particular farming communities, which may be at a distance of up to 15 kilometres from the shore.¹⁹

It was probably in these fishing camps and neighbouring villages that partner exchange reached the frequency required to raise HIV to the epidemic levels elsewhere found only in the urban environments of Kinshasa or Kigali. Fifteen years later researchers studied such a fishing community in Masaka district. Its men had on average one new sexual partner every twelve days. Some 41 per cent of their partners were regular and 59 per cent casual; 85 per cent were contacts within the village, 8 per cent in other fishing villages, and 6 per cent in the nearby trading town. The village women, in turn, had 90 per cent of their sexual contacts with other villagers and 42 per cent with casual, paying clients. Such promiscuity was highly localised, so that HIV prevalence in different parishes of the district in the mid 1990s was to range from 4 per cent to 20 per cent. ‘It is our mating patterns that are finishing us off,’ a researcher was told.²⁰

Although this epidemic began in the countryside, the difficulty of transmitting HIV makes it likely that it would have died away if it had not been carried to more open sexual networks in trading centres, the capital, and eventually the entire East African region. The researchers in Masaka found surprisingly little sexual exchange between village and town, but they did find that sexual activity varied enormously between individuals.²¹ It was perhaps hyperactive and mobile individuals who transmitted HIV to the main-road trading centres where it next flourished. In the Kagera region, for example, the virus appears to have been carried from border trading posts to inland commercial centres like Kamachumu, long a focus of coffee marketing and politics. Thence it spread to the regional capital, Bukoba. By 1987 prevalence among those aged 15–24 was 24.2 per cent in Bukoba town (reaching 42 per cent in its lowest-status section) and 10 per cent in the neighbouring Bukoba and Muleba rural districts.²² Once the virus was established in trading towns, workers carried it back to hitherto unaffected villages. In the Kagera village studied by Gabriel Rugalema, for example, Aids was introduced in 1987 by ‘a woman with an unstable marriage who worked part-time as a commercial sex worker in Rwamishenye (a suburb of Bukoba town). She came back to the village after she had been weakened by infections and died a few weeks later.’ Another 18 women and 41 men died there during the next nine years:

A majority of the men who died were involved in off-farm income generation, particularly those who had worked as itinerant traders. Others included carpenters, masons, and casual labourers . . . Only six of the deceased men could be strictly classified as full-time farmers. . . . As for the women, the majority of the deceased were, as may be expected, full-time farmers.²³

In Rakai district, similarly, a computer simulation suggests that the annual incidence of new infections among people aged 15–24 peaked in 1987 at about 8.3 per cent.²⁴ Two years later, prevalence among men and women aged over 13 varied from 26 and 47 per cent respectively in main-road trading centres to 22 and 29 per cent in local trading village and 8 and 9 per cent in agricultural villages. In 1990–2, 31 per cent of all households in Rakai district contained an infected member. The worst impact was in the truck-stop towns along the trans-African highway between Kampala and Kigali, notably Lyantonde, where HIV was found in 67 per cent of the bar girls tested in 1986 and in 53 per cent of the entire adult population in 1989.²⁵

The prominence of the trans-African highway was one indication that the epidemic had by the mid 1980s spread far beyond the west lake region. Three categories of mobile men appear especially to have carried it. One was the military: General Amin’s soldiers retreating from the infected border region in 1978–9, Tanzanian troops pursuing them through western and northern Uganda, and Ugandan forces seeking to repress rebellion in the north and east during the 1980s. The northern Gulu district, the chief source of Amin’s troops, recorded 15 per cent prevalence among pregnant women in 1987 and probably became the main route by which HIV entered the southern Sudan and was carried northwards by soldiers and refugees to Khartoum, where in 1998–9 nearly half of those infected had the D subtype found in Uganda and DR Congo.²⁶ The western Ugandan districts of Kabale, Kasese, and Kabarole, prominent in early Aids returns, may also have been infected initially by rival armies. Military actions during the 1980s in Luweero and Soroti districts, further east, were probably important in spreading the disease there.²⁷

A second group carrying the virus were long-distance drivers who infected or were infected by bar girls at their overnight stops in towns like Lyantonde. One study of 68 drivers in Kampala in 1986 reported that 35 per cent already had HIV.²⁸ The third occupational group, with a more diffuse and less certain impact, were migrant labourers carrying the disease to rural homes. ‘With the AIDS pandemic,’ a hospital in the remote south-west of Uganda reported in 1991, ‘it is still the returnees to Bufumbira that introduce this deadly disease into the population which otherwise knows no promiscuity. Among the returnees are also counted the taxi drivers and the long-distance truck drivers.’²⁹

Kampala held a special position. In retrospect, its main prison may have held cases as early as 1979 or 1980, when patients with aggressive Kaposi’s sarcoma also appeared in the main Mulago Hospital, soon followed by others with the chronic diarrhoea and wasting of Slim disease.³⁰ Nobody linked these infections to the emerging Aids epidemic elsewhere in the world until late in 1984, by which time HIV was already entrenched in the city and spreading rapidly. ‘It all started as a rumour,’ the chief epidemiologist later reflected. ‘Then we found we were dealing with a disease. Then we realised that it was an epidemic. And, now we have accepted it as a tragedy.’³¹ Studies of prevalence among pregnant women in Kampala showed 11 per cent in 1985, 14 per cent in 1986, 24 per cent in 1987 – then the highest figure in the world outside Kigali – and a peak of over 30 per cent in 1989.³² Notably, however, Kampala’s epidemic was not focused on a core group of sex workers and their clients, in contrast to other East African cities. There was little association between HIV infection and commercial sex, which was unorganised, diverse, illegal, and impossible to distinguish from other sexual relationships involving gifts.³³ Instead, Kampala’s sexual pattern was closer to Kinshasa’s, with more young women than young men, sexual debut at an average age of fourteen in Uganda generally in 1989, 69 per cent of men and 74 per cent of women aged 15–19 having sexual experience, a rising age at marriage, and many young women whose dependence on gifts from male lovers had been accentuated by the economic disorder of the 1970s and 1980s.³⁴ It was a pattern vulnerable to HIV but capable of change.

Although reports of a novel disease in Rakai reached the authorities in 1982–3, Uganda was then in the midst of civil war and no action was taken until Lwegaba’s report coincided late in 1984 with laboratory evidence that patients at Mulago Hospital with Kaposi’s sarcoma were infected with HIV. Milton Obote’s government, then in power, ordered an investigation. A team visited Masaka, conducted examinations at Mulago, and concluded that Slim was ‘part of the spectrum’ of Aids, although with opportunistic symptoms specific to East Africa. Ruling out transmission by casual or indirect means, the researchers blamed heterosexual promiscuity, perinatal transmission, and blood transfusion, estimating that Mulago Hospital might be creating two new cases each day. HIV-positive patients at Mulago reported on average twice as many sexual partners as HIV-negative patients. Another risk factor was a sexually transmitted infection, especially genital ulcer disease.³⁵

Uganda’s HIV epidemic appears to have peaked in 1991, when 21.1 per cent of women attending antenatal clinics tested positive and some 1,200,000 people were thought to be infected.³⁶ By then the virus had reached almost all parts of East Africa. In Tanzania, the area first affected after Kagera was probably Dar es Salaam. An expatriate may have contracted the disease there as early as 1980, but the first firm evidence was a prevalence of nearly 2 per cent in stored blood collected from pregnant women and blood donors in 1984–5. Thereafter antenatal prevalence in the city rose to 8.9 per cent in 1989 and 14.8 per cent in 1997.³⁷ The disease was probably introduced from Kagera, perhaps by returning soldiers but more probably by Haya sex workers and bar girls from the region, who had been prominent throughout East Africa since the interwar period, driven perhaps by male control of land and income in a highly commercialised region. By 1986, 29 per cent of Dar es Salaam’s bar girls had HIV, with a prevalence of 35 per cent among the 33 per cent of them who came from Kagera. Two years later, 60 per cent of notified Aids patients in Dar es Salaam originated from Kagera, many of them no doubt people seeking treatment. Of Tanzania’s first 212 notified Aids cases, 60 per cent of males and 46 per cent of females said that they were heterosexually promiscuous.³⁸ Yet this initial social profile was soon obliterated by the epidemic’s expansion. When women at family planning clinics in Dar es Salaam were surveyed in 1991–2, there was still a positive association between HIV infection and number of sexual partners, but even infected women had a median of only two partners within the previous five years, while married women claiming fidelity to husbands had a significantly greater risk of infection if the husband had not been faithful, a risk that increased with the woman’s own education and her partner’s.³⁹

Dar es Salaam was a thousand kilometres from Kagera and almost as remote from Tanzania’s other borders, yet by August 1986, less than two years after HIV was recognised in the capital, its main hospital had admitted cases from each of mainland Tanzania’s twenty regions. Some were probably infected from Dar es Salaam. In 1988 the highest prevalence, after Kagera, was in Iringa region on the Tanzam road linking the capital to Zambia.⁴⁰ Other areas, by contrast, acquired HIV by cross-border contact. In the south-western Mbeya region, for example, an explosive HIV and tuberculosis epidemic between 1986 and 1994 was caused by the C subtype of HIV-1, probably introduced from Zambia to the south and most prevalent at the border and in urban and roadside locations.⁴¹ Mwanza region, south of Lake Victoria, was probably infected from Kagera, but Mara region, on the eastern shore of the lake, appears to have shared the severe epidemic in the neighbouring Nyanza province of Kenya. In the Kilimanjaro and Arusha regions of northern Tanzania the disease was blamed on young, mobile traders returning from Kagera, Dar es Salaam, and Kenya. As everywhere in the continent, the epidemic there took its shape from the structure of the commercial economy, with a focus among urban bar girls and sex workers, high infection among young adults driven from fertile mountainsides by land scarcity, and prevalence declining as the disease radiated out into the countryside. In Arusha region in 1992, for example, adult infection was 10.7 per cent in the poorer parts of the regional capital, 5.2 per cent in the wealthier parts, 2.2 per cent in semi-urban areas, and 1.6 per cent in the countryside, where at this time the disease was still seen as a complaint of despised urban aliens.⁴² Because HIV entered Tanzania from all directions, the country had an unusual diversity of subtypes and unique recombinant forms.⁴³ Twenty-five years after its first appearance in Kagera the disease was still spreading into remote parts of the country.

While the link from the west lake epidemic to Dar es Salaam was strong, that to Nairobi and the Kenyan epidemic is no more than probable. Kenya’s first Aids cases were concentrated in three locations: Mombasa on the coast, Nairobi in the centre, and the Nyanza province on the eastern shore of Lake Victoria. Any of these may have infected the others, or each may have been infected separately. If HIV reached the two cities directly from west of the lake, the main link, as in Dar es Salaam, was probably women from Kagera prominent in low-status sex work in Kenya since the interwar period. Of 418 women of this kind studied in Nairobi in 1985, 358 were Tanzanians and 37 Ugandans.⁴⁴ Blood specimens tested retrospectively showed that even in 1981 some 4 per cent of the city’s sex workers were infected, a proportion that grew exponentially to more than 85 per cent in 1986. Of men with genital ulcer disease attending a Nairobi clinic, 3 per cent had HIV in 1981 and 15 per cent in 1985, leaving doubt whether women or men were first infected. In 1985, 2 per cent of women at antenatal clinics also tested positive, showing that infection was spreading to the general population.⁴⁵ That was the year when the Kenyan authorities belatedly admitted that the disease was present.

For epidemiologists, HIV in Nairobi was a classic example of an epidemic rapidly transmitted within a core group and then passed on by a bridging group – the sex workers’ clients – to the general population. This happened in Nairobi, as not in Kampala, partly because the Ugandan epidemic began in the countryside and partly because of differences between the cities. In 1979 Nairobi’s 827,775 people included 138 males for every 100 females, with an even larger imbalance among adults. At least half its employed men had no wife in the city.⁴⁶ Wealth and poverty were sharply juxtaposed and women with little education seldom found formal jobs. The result was an exceptionally overt, mercenary style of commercial sex, especially in the Pumwani red-light district, where a community of over a thousand sex workers, many from Kagera, sat outside their rooms waiting for brief encounters with working men at a price of 30–50 US cents. Each averaged nearly a thousand partners a year, working only by day because the night was too dangerous. Some 42 per cent had genital ulcer disease.⁴⁷ Study of their clients in 1986–7, when the epidemic peaked, found that 8 per cent contracted HIV from them and that 96 per cent of infected clients were either uncircumcised or had genital ulcer disease or both. Five years later, 76 per cent of women in Nairobi seeking treatment for a sexually transmitted disease reported only one partner during the previous three months and had presumably been infected by him, indicating the potential for transmission to the general population. HIV prevalence at Nairobi’s antenatal clinics may have peaked in 1994 at about 17 per cent. Four years later over 40 per cent of Kenya’s new HIV infections were thought to come through commercial sex.⁴⁸

The sex workers themselves suffered terribly. Nearly half of those hitherto uninfected contracted HIV each year. They then generally developed Aids within about half the normal time, perhaps owing to multiple infection or other sexually transmitted diseases.⁴⁹ Their danger was discovered almost accidentally in 1985 during a preliminary survey of sexually transmitted diseases. When astonished researchers told sex workers that two-thirds of the 60 tested had HIV, they met ‘stunned silence’. Only five wanted to know their personal status, although most quickly adopted the free condoms pressed upon them. ‘When one gets beyond the initial prejudices and stereotypes,’ the organisers wrote, ‘one finds the prostitute knowingly risking AIDS, sacrificing her own hopes for the sake of her children or brothers and sisters.’⁵⁰

The explosive epidemic in Nairobi almost monopolised attention in Kenya, so that little is known of HIV elsewhere during its first decade. Perhaps misleadingly, the coast region reported three times as many Aids cases as Nairobi in 1991, the great majority no doubt in Mombasa, where 54 per cent of 3,628 sex workers tested positive between 1993 and 1997 and adult prevalence in 2000 was 10.8 per cent.⁵¹ Elsewhere prevalence during the early 1990s was relatively low, except in towns along the trans-African highway between Nairobi and the Ugandan border. In 1993 both Nakuru and Busia reported higher antenatal prevalence than either Nairobi or Mombasa. From the mid 1990s there was also rapid growth in the Central and Eastern provinces around Nairobi. Kenya’s adult infection rate probably peaked around 1998, officially at 13.9 per cent although the true mark may have been substantially lower.⁵²

Kenya’s anomaly was the Nyanza province bordering Lake Victoria, which experienced an explosive epidemic that is perhaps the least understood in Africa. The earliest infections may have come across the lake soon after the epidemic began on its western shore, for between 1986 and 1993 Nyanza reported 15,605 Aids cases – 31 per cent of all Kenya’s cases – implying widespread HIV prevalence in the early 1980s at least. By 1993, prevalence at antenatal clinics in Kisumu, the regional capital, was 20 per cent and rising quickly.⁵³ In the absence of detailed analysis, the best explanation of this epidemic suggests a combination of circumstances fostering disease elsewhere but seldom joined in one place. One was participation in lakeshore fishing culture. ‘The beaches attract a continual inflow of people,’ it was reported: ‘young men in pursuit of an easy cash income and women following the men. They live outside the traditional social structure and subsistence farming households, and drinking, casual sex, theft, HIV/AIDS and high death rates among young men are common.’ Nearly half the adults in these areas may have been infected by the early 2000s.⁵⁴ Equally vulnerable were young people with casual jobs on sugar plantations and especially on the fringes of the transport industry, for Nyanza straddled the trans-African highway and had its own motor transport network. Its dense rural population, closely linked to the urban focus of infection in Kisumu, bred rural prevalence levels among adults reaching 30 or 40 per cent in the early 2000s, while scarcity of land and lack of rural opportunity perpetuated migration to Kampala, Nairobi, and workplaces throughout Kenya, where Nyanza people often had exceptionally high rates of HIV.⁵⁵

The social organisation of the Luo people also contributed to the epidemic. One study attributed over half their infection to the fact that some 90 per cent of Luo men, unlike most Kenyans, were not circumcised.⁵⁶ Their society was strongly patriarchal. In interviews at clinics in Kisumu in 2000, with Luo forming 81 per cent of those questioned, men reported unprotected sex with an average of 11.2 partners, women with 2.5. It is not clear whether these women included sex workers, but they numbered an estimated 1,400 in Kisumu in 1997–8 and 75 per cent of them were HIV-positive.⁵⁷ Many were probably divorced or separated women with few other opportunities in Luo society. Luo themselves saw the epidemic as only the culmination of a century of economic decline and social disintegration, focusing particular attention on their custom of inheriting widows (and hence, supposedly, the virus that killed their husbands) and on alleged youthful promiscuity. One study in the rural Asembo area in 2004 showed that 33 per cent of boys and 22 per cent of girls under fourteen years of age claimed sexual experience, which had probably been common among youths in the past but had taken non-penetrative forms. A survey of women in Kisumu in the late 1990s found HIV infection only among those who had engaged in premarital sex.⁵⁸ Female prevalence there rose from 8 per cent at age 15 to 29 per cent at age 17, at which age only 2 per cent of men were infected. Of every five people with HIV, three were women.⁵⁹ The connection between gender inequality, sexual behaviour, and vulnerability could scarcely have been stronger.

Except, perhaps, in Ethiopia. In its origins the Ethiopian epidemic differed from those elsewhere in eastern Africa, but in most other respects it was, despite the country’s distinctive history, surprisingly similar, especially in the unsuspected early spread of heterosexual infection arising from sexual exploitation of women. The problem in Ethiopia, however, is not, as in Nyanza, why an extensive epidemic took place, but rather, as in pre-genocide Rwanda, why the epidemic was not more extensive. This may seem paradoxical, for in the early 2000s about 1,500,000 Ethiopians had HIV. Yet that implied a prevalence in those aged 15–49 of 4.4 per cent, only half the proportion in Tanzania and two-thirds of that in Kenya.⁶⁰

One reason restricting the epidemic was that HIV reached Ethiopia somewhat later than the other eastern African countries. The first two cases were diagnosed in Addis Ababa in 1986. Retrospective tests on stored blood revealed one case in 1984 and another in 1985, but none in earlier specimens. Analysis of the diversification of the virus suggested that it had arrived in 1983. The virus itself, introduced at least twice, was subtype C of HIV-1, in contrast to the A and D subtypes dominant in East Africa. How the subtype mainly found in southern Africa and India also reached Ethiopia is unknown, but its complete domination of the epidemic – in contrast to the diversity of subtypes in Tanzania – suggests not only Ethiopia’s isolation but a rapid saturation of a core group of vulnerable people from whom the infection spread to the wider population.⁶¹

The core group were the sex workers of Addis Ababa and other major towns, together with their habitual clients. Founded in 1886 on the model of a military camp, the capital was a sprawling jumble of permanent buildings and the squatter shacks in which over four-fifths of its nearly two million people lived. Women were a majority of the population, especially in the younger age ranges, for Ethiopian women married very young, divorce was common, and there was little place for unmarried women in the countryside. In the town, such women survived chiefly by informal activities, of which commercial sex was one of the most important. In 1973–4 an Ethiopian sociologist reckoned that some 27,000 women worked in bars, the chief meeting places for the city’s men. An official survey in 1982 identified 15,900 full-time sex workers in the city. A less official one, seven years later, estimated 24,825, excluding streetwalkers and women working from their own rooms, adding that 55 per cent had only one or fewer partners per week.⁶² Divorce, disagreement with parents, and lack of money to continue schooling were reasons often given for entering commercial sex. Major provincial towns had smaller but similar groups of sex workers.

Commercial sex had a role in Ethiopian urban culture similar to that in Kigali. Female virginity at marriage was vital to respectable families, if perhaps less so to their daughters than in the past, partly because marriage ages were rising with education. Men, by contrast, suffered little inhibition on sexual experimentation and on average (in 2000) married seven years later than their wives. Given this imbalance, as in Kigali, young men commonly had their first experience with sex workers and up to half continued to frequent them thereafter. Early in the epidemic most of these sexual encounters were unprotected, for Ethiopians were unfamiliar with condoms and hostile to them.⁶³ Sexually transmitted diseases liable to facilitate HIV transmission were common, especially among sex workers. A study in Addis Ababa in the early 1990s found that only 9 per cent of women in their first marriage and 1 per cent of sex workers had no serological evidence of such a disease, while 33 per cent and 46 per cent, respectively, were infected with HSV-2, which caused genital ulcers and particular susceptibility to HIV. Moreover, Ethiopia’s health services were slender even by African standards, taking only 0.4 per cent of the national budget in 1999 and providing fewer than 20 per cent of pregnant women with antenatal care, as against an average of over 60 per cent in sub-Saharan Africa.⁶⁴

HIV first became established in Addis Ababa among sex workers during an explosive epidemic in the late 1980s. In 1987, 5.9 per cent of them tested positive; by 1990 the figure had risen to 54.2 per cent. Prevalence was especially high in city centre brothels. By contrast, in 1989 only 4.6 per cent of the capital’s pregnant women were infected.⁶⁵ Other places of very high prevalence among sex workers at this date were the trucking towns of Dessie, Nazareth, Mekele, Bahr-Dar, and Gonder on roads radiating outwards from Addis Ababa. In the far north, however, the disease was still rare, although it had penetrated to all parts of the country. Study of 23 towns in 1988 showed an average prevalence of 17 per cent among sex workers, 13 per cent among long-distance truck drivers, but only 3.7 per cent among blood donors (who broadly represented the general population).⁶⁶ Among the latter, rapid epidemic growth began three or four years later than among sex workers, the annual incidence of new urban infections peaking in 1991 at about 2.7 per cent. Prevalence among antenatal women in Addis Ababa rose from 4.6 per cent in 1989 to 11.2 per cent in 1992–3, reaching its likely peak of 21.2 per cent in 1995.⁶⁷

At the same time, the ratio of infected men to infected women in the capital fell from 3.7:1 in 1988 to 1.5:1 in 1994, suggesting that an epidemic that had begun among a core group had spread to the general population. In a study of 2,526 factory and estate workers in and around Addis Ababa in 1994, HIV infection in men was strongly associated with reported sexual behaviour and past history of syphilis, but in women it was associated with sociodemographic characteristics (low income, low education, and living alone) rather than sexual behaviour. Moreover, the burden fell increasingly on young women. In 1995, antenatal prevalence in Addis Ababa was 23.7 per cent among women aged 15–24, 17.7 per cent among those of 25–34, and 11.1 per cent among older women. In Dire Dawa, a railway town east of the capital, 57 per cent of all infected women in 1999 were aged 15–24.⁶⁸

Ethiopia’s urban epidemic ceased to expand during the mid 1990s, although numerous new infections continued to compensate for the rising number of deaths. The missing element in the story, however, was expansion to the countryside, for the remarkable point about Ethiopia – in contrast, say, to Nyanza – was how little impact the disease had made in rural areas, where estimated adult prevalence was 0.3 per cent in 1990 and 0.8 per cent in 1995. This was partly misleading, for such was the predominance of the countryside in Ethiopia – 83 per cent of the population in 1999 – that rural infections overtook urban from 1997. Yet rural prevalence in 2000 was still only an estimated 1.9 per cent. It was highest in the central Amhara region, but in the remote Southern Nations Nationalities and Peoples Region, at that date, only 37 per cent of women had even heard of Aids, although the impact grew rapidly thereafter.⁶⁹ Rural people, there and elsewhere, blamed townsmen and foreigners for the disease: ‘We Hamar don’t have cars with which to reach America. We don’t go to England, to gal [highland] country, to Germany, and going there, we don’t come back bringing illness. It comes to us by foot.’⁷⁰

As in Rwanda and Burundi, it is difficult to explain the weakness of urban–rural transmission of HIV in Ethiopia during the 1990s. One element may have been the dispersed pattern of rural settlement that limited interaction. Studies of the extent to which farmers frequented sex workers in market towns found inexplicably varied proportions.⁷¹ As in Rwanda, occasional visits might do little to spread a virus so difficult to transmit, especially in a culture with near-universal male circumcision. The most detailed rural study, of a Muslim area in eastern Hararghe, concluded that it was protected from infection by its Muslim social order and its lack of exposure to high-prevalence urban groups.⁷² Perhaps this last point was the most important. The HIV/Aids epidemic throughout eastern Africa had been shaped by the network of communication provided by commercial economies. Vigorous around Lake Victoria and along the trans-African highway, they were less integrated in Rwanda and Burundi or the emptiness of central Tanzania. The particular weakness of its commercial economy had shaped much of Ethiopia’s modern history, notably its uncompleted revolution. Now the same circumstances helped to protect its countryside against infection.