correspondence Helen Weiss, MRC Tropical Epidemiology Group, Infectious Disease Epidemiology Unit, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT. E-mail: firstname.lastname@example.org
Summary The HIV pandemic continues to evolve, in both magnitude and diversity. In this paper, we briefly review the global epidemiology of HIV/AIDS, reflecting on the differences by region, and the challenges posed by the evolving epidemics in terms of prevention and surveillance. Despite the reduction in numbers of new AIDS cases in the US and Western Europe due to advances in treatment, a constant number of new HIV infections persists every year, with evidence that in some settings high-risk behaviour has increased, indicating failure in primary prevention. However, the vast majority of new infections still occur in developing countries. The two giants, India and China, are set to determine Asia's ultimate contribution to the pandemic, but currently it is Africa that remains in the eye of the storm, with rapidly increasing epidemics in many countries in the south and east of the continent impacting on all levels of society.
After two decades of the HIV/AIDS pandemic, the natural history and transmission of HIV are well understood, and its global epidemiology is generally so well-characterized that one may ask what more epidemiology can contribute. The answer is: a great deal, if we view epidemiology as a science for public health action. In this paper, we briefly review the global epidemiology of HIV/AIDS, with a particular focus on two extremes, the United States and sub-Saharan Africa, and reflect on the challenges posed by the evolving epidemics in terms of prevention and surveillance.
UNAIDS estimates that 5.6 million new HIV infections and 2.6 million AIDS deaths occurred in 1999, and almost 34 million adults and children worldwide are currently living with HIV or AIDS, the great majority of them in the developing world ( UNAIDS/WHO 1999). The fact that there are disparities in HIV epidemiology between industrialized and developing regions is unsurprising given the substantial differences within and between the industrialized nations. For example, Figure 1 shows the incidence of AIDS per million for the United States and Western Europe. The rate in the United States in 1996 was 8 times higher than in the United Kingdom or Germany, 3 times higher than in France or Switzerland, and 1.2 times higher than in Spain.
HIV/AIDS epidemics in the United States
In the United States, persons with advanced HIV disease invariably have access to medical care, and hence ascertainment of AIDS cases is assumed to be relatively complete. On the maps in Figure 2, each dot represents 30 cumulative cases of AIDS. They illustrate the progression of the epidemic over time and place from an initial small number of cases in cities along the north-east, south-east and west coasts in the early 1980s. There was extension around the original epicentres in the mid-1980s, followed by spread throughout the eastern part of the country by the late 1980s. The south was increasingly involved by the mid-1990s, and the map of today shows that only the mid-west is relatively spared. By June 1999, a cumulative total of 702 748 AIDS cases had been reported, of whom 84% were male. A total of 47 083 cases of AIDS were reported in the period July 1998–June 1999, for an annual incidence of AIDS ranging from a low of 0.9/100 000 in North Dakota to 143/100 000 in Washington DC ( CDC 1999a).
In 1998, 35% of new AIDS cases in the US were in homosexual men, 23% in injecting drug users, and 14% were attributed to heterosexual contact. Important trends recently have been an increasing proportion of cases in women; a decreasing proportion among homosexual men; an increasing proportion attributable to heterosexual contact; and an increasing proportion of cases in the south. Especially striking have been the trends in AIDS incidence by race/ethnicity. In 1998, 48% of new AIDS cases were in African Americans, 31% in non-Hispanic whites, and 20% in Hispanics ( CDC 1999a). The population-based incidence rates of AIDS in African American men in 1998 were 7 times, and in African American women 25 times, higher than in whites. Important differences exist between racial and ethnic groups in the distribution of AIDS cases by exposure category. Among males with AIDS, twice as many whites as African Americans were homosexual; in contrast, the proportions of AIDS cases in African Americans that were injecting drug users or resulted from heterosexual exposure were each three times greater than in whites.
The most striking trend has been the decline in AIDS incidence and deaths that has occurred since 1995 when combination antiretroviral therapy first began to influence the natural history of HIV infection and disease ( Figure 3). The annual number of AIDS deaths increased steadily until 1995, followed by a dramatic decline from 1995 to 1998. In 1998 there were 20% fewer deaths than during 1997; the decrease from 1996 to 1997 was 42% ( CDC 1999a). The number of deaths per quarter has been approximately constant since late 1998. The curve of AIDS incidence shows the effects of two forces: the expansion of the AIDS case definition that was implemented in 1993, and the effect of therapy after 1995. AIDS incidence in 1998 was 11% less than in 1997; and in 1997 was 18% less than in 1996. The number of diagnoses per quarter began to level off during 1998. These trends have been observed for men and women, in all exposure and racial/ethnic groups, and in all geographical regions, although to unequal degrees. For example, AIDS incidence and deaths have declined more rapidly in whites than in African Americans or Hispanics. The numbers of cases of perinatally acquired AIDS cases have decreased by 75% between 1993 and 1998 as a result of the HIV testing of pregnant women and provision of zidovudine.
Implications for surveillance
The fundamental point concerning surveillance trends in adolescents and adults is that AIDS case surveillance no longer provides the same window on the epidemiology of HIV infection, since an AIDS case today represents not only a failure of primary prevention of HIV infection, but also of secondary prevention of HIV disease. This may result from failure to diagnose HIV infection, failure to access care and therapy, failure to adhere to therapy, or failure of therapy itself due to drug resistance or some other cause. We do not have data at present to be certain which of these factors are primarily responsible for the plateauing of AIDS cases and deaths observed since 1998.
In response to the effect of treatment on AIDS trends, the Centers for Disease Control and Prevention (CDC) recommend, in recently published guidelines, that AIDS case surveillance be extended to include the reporting of all persons confidentially diagnosed with HIV, using the same approaches and infrastructure as used for AIDS case surveillance ( CDC 1999b). Specifically, this means HIV-infected persons would be reported confidentially by name to the surveillance system, as has been standard practice for persons with AIDS. Although such AIDS case reporting has been long and well-accepted, named HIV surveillance has been controversial. CDC judge that alternative systems using coded identifiers have not performed adequately. In addition to the population-based reporting of HIV and AIDS cases, CDC intend to use a sampling approach to extend their monitoring of morbidity and use of care; to focus behavioural surveillance on populations at higher risk of HIV and; to target serosurveillance on intensely affected areas, focusing on HIV incidence as prevalence.
Implications for primary prevention
Turning to primary prevention of HIV, the available evidence suggests that in the United States incidence of HIV infection has remained stable for the past few years, and our best estimate is that approximately 40 000 new HIV infections occur each year. The belief that HIV incidence is stable (which cannot be regarded as a public health success) comes from several sources. First, HIV case reports from states that introduced HIV case reporting before 1993 are essentially constant. In persons younger than 25 years, AIDS cases necessarily reflect that constant subset of recent HIV infections where the absence of treatment progressed quickly to disease: such trends also are stable. Directly measured HIV incidence in repeat testers such as homosexual men and individuals attending STD clinics continue to be high. Finally, direct measurement of incidence is now possible using the ‘detuned assay’, a dual-testing strategy using a sensitive and less sensitive test in order to detect recent infection ( Janssen et al. 1998 ). Data from San Francisco over a 10-year period show an essentially constant HIV incidence in homosexual men attending STD clinics.
There is now worrying evidence that in some settings high-risk behaviour has increased. For example, an increasing proportion of gonorrhoea cases nationally is occurring in homosexual men. Men who have sex with men in San Francisco have shown an increased rate of rectal gonorrhoea in parallel with less consistent condom use and unprotected rectal sex ( CDC 1999c), and a large outbreak of syphilis in homosexual men has been reported in Seattle ( CDC 1999d). In conclusion, with the exception of mother-to-child transmission, as best as we can judge, we have not succeeded in reducing HIV incidence during the past few years, and therapeutic advances have deflected attention from the underlying continuing transmission of HIV. This implies failure of primary prevention to curb the spread of HIV at a time when tracking the epidemic is more difficult than ever.
HIV/AIDS epidemics in the European Union
By June 1999, the 15 countries of the European Union had reported a cumulative total of around 206 000 AIDS cases, less than one third of the total in the US, which has a slightly smaller population. As in all industrialized regions, AIDS incidence has dropped with advances in treatment since 1996, although prevalence is increasing due to a sustained number of newly diagnosed HIV infections ( CDSC 1999; European Centre for the Epidemiological Monitoring of AIDS 1999). The epidemics in Europe are heterogenous, with injecting drug use continuing to be the predominant mode of transmission in several southern European countries, while sexual transmission dominates in northern Europe. The numbers of new AIDS cases due to male-to-male sexual transmission has fallen by 24% overall, and heterosexual contact has become the predominant mode of transmission among new cases in several countries including France, Belgium, Sweden, Norway and the United Kingdom ( European Centre for the Epidemiological Monitoring of AIDS 1999; CDSC 2000). Advances in perinatal HIV prevention in Europe resulted in less than 100 cases of AIDS transmission from mother-to-child in 1998.
HIV epidemics in Eastern Europe and Central Asia
The epidemic spread of HIV in Eastern Europe started around 1995, and is increasingly dynamic and unstable with the total number of HIV infections in this region estimated to have risen by a third during 1999 to approximately 360 000 ( UNAIDS/WHO 1999). The most heavily affected parts of the former Soviet Union are Ukraine and Russia, which accounted for the great majority of reported HIV infections in the newly independent states in 1998. In Russia, almost 5000 new cases, half of them in Moscow, were reported in the first 6 months of 1999, and infections reported in 1999 constitute approximately half of reported cases since the epidemic began ( European Centre for the Epidemiological Monitoring of AIDS 1999). The driving force for the whole region is injecting drug use. Nevertheless, in some countries such as Russia, modes of transmission are heterogeneous, including drug use, male-to-male sex and heterosexual contact. In addition, major epidemics of tuberculosis and syphilis have occurred against a background of political and economic instability, increased mobility, drug use, and degradation of health infrastructure. These countries of Eastern Europe and Central Asia constitute a region where the future course of the HIV epidemic is most unpredictable.
HIV epidemics in Asia
Moving east, the most heavily affected countries of south-east Asia are Cambodia, Thailand, and Myanmar, with injecting drug use and commercial sex driving HIV transmission. The epidemic in Thailand has been best documented, showing a rapid spread in the late 1980s, initially among injecting drug users and commercial sex workers, and subsequently into the general population, leading to prevalence rates of up to 10% among male military recruits and up to 6.4% among antenatal clinic attenders by 1993–94. However, rates in all population groups subsequently declined following a comprehensive control programme established by the Thai authorities, which included enforced use of condoms in establishments used by sex workers and a mass advertising campaign ( Rojanapithayakorn & Hanenberg 1996). There remain differences in prevalence within Thailand, for example for pregnant women, among whom prevalence exceeds 5% in the north but is approximately 2% in Bangkok in the south ( US Bureau of the Census 1999).
Vulnerable groups of injecting drug users and commercial sex workers exist in all countries in the region, and prevalence of infection among injecting drug users remains high, with estimates in Thailand ranging from 20 to 45%, and in Rangoon and Mandalay in Myanmar from 50 to 85% in 1992–93. Rates of HIV infection among sex workers in Phnom Penh in Cambodia exceed 50% ( Oshige et al. 2000 ), and the estimated prevalence among all adults is 3–7% among women and 6–16% in male police and military personnel ( Ryan et al. 1998 ). Despite the infection rates in these high-risk groups, rates of infection in the general population have remained low (< 0.5% among adults aged 15–49) in several countries with large populations including China, Malaysia, and the Philippines ( WHO/WPRO 1999).
The combined populations of India and China exceed two billion, accounting for more than one third of all of the world's citizens, and the course of epidemics in these two countries is key to the future of the pandemic. Available epidemiological data are regrettably weak. Overall it is estimated that around 4 million Indians are HIV-infected, but the epidemic is extremely heterogeneous – for example, in Mumbai, Pune and Tamil Nadu state, HIV prevalences of 2–4% have been reported in pregnant women, contrasting with Calcutta and Delhi where prevalence has remained low (0–0.8%). Injecting drug use is the predominant mode of transmission in the north-eastern states near Myanmar, where prevalence among drug users exceeds 70%. In China, appreciable levels of HIV infection were first noted in Yunnan Province bordering the Golden Triangle where Myanmar, Thailand and Laos meet. The prevalence of HIV infection in injecting drug users here climbed rapidly to 70% or more in the early 1990s, and HIV first emerged in sex workers in this region. At least 400 000 persons are believed to be living with HIV in China, and the epidemic is currently concentrated among injecting drug users.
HIV epidemics in Latin America and the Caribbean
The epidemic in Latin America was similar initially to that in North America and Europe, with most cases among homosexual men and injecting drug users. Systematic surveillance in the region is limited and the picture is complex, with male-to-male transmission predominating in some countries (Mexico, Chile, Cuba), while in others (Argentina and Brazil), injecting drug use accounts for about half of all infections. However, in almost all countries, there has been a rapid increase in cases attributed to heterosexual transmission, with a corresponding increase in the proportion of infections occurring among women. For the region as a whole, about 20% of HIV-infected adults are women, with the highest proportion in Brazil (31%), where AIDS has been the leading cause of death among women aged 20–34 since 1992 in Sao Paolo ( Monitoring the AIDS Epidemic (MAP) 1998). HIV rates in pregnant women are generally low, although there are estimates of 4% in Honduras, and 3% in urban areas of Brazil and Argentina.
Heterosexual transmission generally predominates in the Caribbean, where overall prevalence is nearly 2% – higher than anywhere apart from Sub-Saharan Africa. The worst affected country is Haiti, where 8% of pregnant women are estimated to be HIV-infected.
HIV epidemics in Africa
In the developing world, antiretroviral treatment is largely inaccessible and so we are still witnessing the ‘natural history’ of the AIDS epidemic. While the two giants, India and China, will determine Asia's ultimate contribution to the pandemic, it is Africa that remains in the eye of the storm. With a population of about 600 million (approximately 10% of the world total), sub-Saharan Africa accounts for over two-thirds of the world's HIV-infected persons, and 80% of the world's HIV-infected women and children. In 1999 approximately 2.6 million AIDS deaths occurred globally, at least 2 million of them in sub-Saharan Africa. In sentinel populations of pregnant women, HIV prevalence in some parts of southern Africa, such as in Francistown, Botswana, now exceeds 40%. Large differences exist in the severity of HIV/AIDS, the epidemic being concentrated in the eastern and southern parts of the continent. While prevalence has stabilized in some areas, a stable prevalence implies a high incidence to replace losses due to mortality, and Africa continues to suffer the highest number of new infections anywhere in the world. The one country which has seen a decrease in HIV prevalence is Uganda, where HIV prevention campaigns since the early 1990s have successfully encouraged many young people to delay the onset of sexual activity and increase condom use ( Asiimew-Okiror et al. 1997 ). The country is now seeing results – whereas infection rates had reached highs of 40% among pregnant women in cities, prevalence in this group is currently around 15%.
Explanations for heterogeneities in the epidemic in different parts of Africa must lie in the interaction between the virus, the human host, and the broader environment. The absence of a pandemic of HIV-2, which has essentially remained concentrated in West Africa, is likely to be related to lower transmissibility compared with HIV-1. We have no definitive evidence that important biological differences exist among the various subtypes of HIV-1 in pathogenicity, infectiousness, or predisposition for specific modes of transmission ( Hu et al. 1999 ). Subtype A predominates in West Africa; A and D in East Africa, and C in Southern Africa. A recent multicentre comparative study has re-emphasized the role of genital herpes and lack of male circumcision as two of the factors associated with high HIV seroprevalence ( Buvéet al. 1999 ).
Public health impact in sub-Saharan Africa
The effects of the epidemic in Africa are obvious across society. HIV/AIDS dominates health structures, rapidly emerging as the leading adult cause of death when HIV prevalence exceeds a few percent. Across the continent, hospitals are filled with wasted people. In the shadow of HIV is an epidemic of tuberculosis, the commonest AIDS-associated illness. Rates of mother-to-child transmission are higher in Africa than in most industrialized countries, largely because of prolonged breast feeding, and infant and child mortality have risen steeply. Increased paediatric and adult mortality results in massive reductions in life expectancy, erasing the gains in health and child survival that have occurred over our lifetimes. The disability and death of young and middle-aged adults removes families' bread-winners, providers of school fees and carers, as well as societies' teachers, health professionals and administrators. A new social group are the AIDS orphans, children who have lost one or both parents to HIV/AIDS, now some 8 million strong in Africa, raising the spectre of a population the size of a small European country of disenfranchised persons growing up with a bleak future. Associated phenomena are child-headed households and elderly grandparents caring for different sets of grandchildren.
Although considerable, the demographic and economic impacts are more obvious at the local than the macro level. High fertility rates mean it is unlikely that the population of any country will decline, although population increase is being significantly curtailed. At the level of the local community and especially the family, however, missing people are only too evident. Similarly, the economic effects of the epidemic may be hidden at societal level, but vivid to the family which becomes suddenly destitute.
The impact needs to be viewed in its historical and political context. 20–25 million Africans currently living with HIV will die prematurely. For comparison, just over one million Americans have died serving their country in all of America's wars; 800 000 Rwandans died in 100 days of genocide in 1994; from the 14th to 18th centuries, plague is estimated to have killed 25 million people worldwide, approximately the number of Africans currently waiting to die from HIV. HIV/AIDS is clearly Africa's greatest social catastrophe of the 20th century, its greatest calamity since the slave trade. The impact and duration of the epidemic in the 21st century remain speculative.
The gap in the HIV/AIDS epidemic between the industrialized and developing worlds continues to widen, compounded now by the widespread use of antiretroviral therapies reducing AIDS incidence and mortality where such treatment is affordable. Challenges in industrialized countries include increased difficulty in monitoring the epidemic (in part because of failure to institute uniform HIV surveillance), weakening of primary prevention from fatigue as well as overconfidence in the efficacy of treatment, provision of early diagnosis and therapy to all those who require them, and prevention and care among isolated, including immigrant, communities.
In sharp contrast, HIV continues to spread in most parts of southern Africa, Asia and Latin America, where antiretroviral treatments are generally not available. In many of these countries the epidemiological ‘set point’ for HIV and AIDS has not been reached. The chronicity of the infection suggests the epidemic may last for many decades, perhaps longer than a century, and the impact in developing countries in the 21st century will greatly exceed that seen until now. In the absence of a vaccine, emphasis must be on integrated interventions across all sectors of society ( Table 1). To date, the global response to this historic epidemic has been inadequate in relation to the magnitude of the problem.
Table 1. Currently feasible but underemphasized interventions for limiting HIV/AIDS impact in developing countries