Developing countries find themselves at different stages of the HIV/AIDS epidemic due largely to differences in the timing of the introduction of HIV and in sexual and drug-injecting behavior. In this section, we classify countries according to the level and distribution of HIV infection, which, as we shall see, has a significant influence on the cost-effectiveness of preventive interventions (chapter 3), and on the scope of interventions to mitigate the impact of the epidemic (chapter 4).

Despite the obvious value of a country typology, low-quality and inadequate data and our still-incomplete knowledge of the disease itself make any evaluation of a country’s status very tentative. Because collecting incidence data is very costly, HIV monitoring systems collect data on prevalence. However, prevalence data are rarely collected for representative samples of the population. Our typology is therefore necessarily based on prevalence among frequently monitored groups with presumed high-risk behavior—sex workers, injecting drug users, homosexual and bisexual men, STD patients, and the military—and one frequently monitored group assumed to be at lower risk—pregnant women attending antenatal clinics. Most of these groups present significant sampling problems. For sex workers, injecting drug users, and homosexual and bisexual men, it is often impossible to identify a representative sample. Even the prevalence data for pregnant women, which may be systematically collected, are not usually nationally representative but are confined to women in urban areas who attend certain clinics. Further, because women only become pregnant if they are sexually active and most pregnant women are from younger age groups, women at antenatal clinics cannot be a proxy for the general population. Thus, much of the information about HIV prevalence comes from ad hoc samples, and in some cases the samples are very small.

Notwithstanding these and additional problems that we discuss below, by using available data from research studies and epidemiological surveillance, countries can be classified according to two broad criteria: first, the extent of HIV infection among groups of people often found to engage in high-risk behavior,¹⁷ and, second, whether the infection has spread to populations assumed to practice lower-risk behavior. The typology includes three stages of the HIV/AIDS epidemic:
 

• Nascent: HIV prevalence is less than 5 percent in all known subpopulations presumed to practice high-risk behavior for which information is available. 

• Concentrated: HIV prevalence has surpassed 5 percent in one or more subpopulations presumed to practice high-risk behavior, but prevalence among women attending urban antenatal clinics is still less than 5 percent.

• Generalized: HIV has spread far beyond the original subpopulations with high-risk behavior, which are now heavily infected. Prevalence among women attending urban antenatal clinics is 5 percent or more.

Due to a lack of data, this typology does not address several important factors. In particular, it does not distinguish between countries based on incidence, the rate of new infections. As we have seen, because HIV cannot be cured and lasts many years, prevalence can rise even if incidence is declining. Prevalence data do not reveal whether the number of new infections is increasing, declining, or level, either in specific subpopulations or among the entire population.¹⁸ Moreover, prevalence data are available only for a few specific subpopulations. Prevalence can stabilize in one or more of these subpopulations even as it spreads rapidly through others that are not monitored. These shortcomings highlight the importance to governments of collecting additional data on HIV incidence and prevalence so that policymakers can formulate an effective response (box 2.7). 

Even with much better data on incidence and prevalence, we would still lack sufficient behavioral information to confidently predict the course of the epidemic in a specific country. Surveys of sexual behavior by the World Health Organization’s Global Programme on AIDS in the late 1980s and early 1990s were among the first attempts to measure behavioral risk factors for HIV infection in developing countries (Cleland and Ferry 1995). WHO also studied drug-injecting behavior and the risk of HIV infection in thirteen cities in industrial and developing countries in 1989 (WHO, Program on Substance Abuse 1994). These and more recent studies have increased our knowledge about risk factors in developing countries. Even so, almost two decades into the epidemic only a few geographical areas and a handful of developing countries have been covered. Policymakers in most countries simply do not know how many people engage in commercial sex, casual sex, or injecting drug use; how frequently they do so; or the extent to which they take actions to reduce their risk of contracting HIV. Basic information on the levels of condom use and partner change among the general population is also unavailable. Without this information, it is simply not possible to accurately predict the course of the epidemic.

For these reasons, we cannot predict with certainty in which low-prevalence countries HIV infection will take off or at what level prevalence will stabilize. Some countries where HIV is currently regarded as a minor problem may turn out to resemble eastern Africa, where the virus spread rapidly through groups with high-risk behavior and widely into the general population. In other countries HIV may infiltrate groups with high-risk behavior but never evolve into a generalized epidemic, even without intervention. Or, knowledge of HIV may cause people to adopt less risky behavior, with or without government urging. Countries with a high incidence and prevalence of STDs other than HIV are likely to be particularly vulnerable to a large and rapidly spreading HIV epidemic, because STDs and HIV are spread by the same behavior and STDs enhance HIV transmission. However, since most other STDs can be cured, countries with low STD prevalence may still have behavior patterns that are conducive to the rapid spread of HIV. 

Lacking the information to predict the course of the epidemic, and given the terribly high human and financial costs of HIV/AIDS, it would be prudent to assume the worst and act aggressively to minimize the epidemic as early as possible. The remainder of this chapter uses this typology to outline the state of the epidemic as of mid-1996 in four developing regions. A list of countries and estimates of prevalence in different subpopulations are in table 1 of the statistical appendix to this report.

Africa

Roughly 90 percent of all HIV transmission in Sub-Saharan Africa is by heterosexual sex. HIV has spread rapidly among people with high-risk behavior and widely among those assumed to be at lower risk. Prevalence among urban sex workers exceeds 20 percent in seventeen countries, and is 50 percent or more in nine countries (figure 2.11). Infection rates among women attending antenatal clinics have grown rapidly to high levels in some areas, have stabilized at lower levels in others, and appear to be declining in Kampala, Uganda (figure 2.12). HIV has infected more than 5 percent of women attending urban antenatal clinics in nineteen countries, and in six countries more than 20 percent are infected. An estimated two-thirds of all new cases of mother-to-child transmission worldwide occur in Sub-Saharan Africa (UNAIDS 1996d).

The countries with generalized epidemics include most in eastern, southern, and central Africa, plus Côte d’Ivoire, Benin, Burkina Faso, and Guinea-Bissau in West Africa (figure 2.13). There is often considerable geographic variation in infection levels within countries. Nigeria, which has more than 100 million people and is the region’s most populous country, has areas at all three stages of the epidemic. In more than half of Nigeria’s states the epidemic is concentrated. HIV has spread most widely in Lagos, along the west coast, and in Delta, Plateau, Borno, and Jigawa states, located to the east and northeast. However, in three states—Edo, Niger, and Oyo—the epidemic is still nascent with low prevalence levels, even among subpopulations with high-risk behavior. HIV was detected early in the Democratic Republic of the Congo (formerly Zaire), but in contrast to many eastern and southern African countries, prevalence has stabilized at less than 5 percent on average in urban antenatal clinics (Piot 1994; statistical appendix, table 1). In Uganda, one of the hardest-hit countries in Africa, HIV prevalence among young people has declined (box 2.8).

Most Sub-Saharan African countries face the dual challenge of lowering HIV prevalence—which can happen only over many years—and of coping with the impact of existing high prevalence on the health system and society. Their domestic budgetary resources to accomplish this are quite limited. Countries with nascent epidemics in Sub-Saharan Africa—Cape Verde, Madagascar, Mauritania, Mauritius, and Somalia—have a unique opportunity to intervene early and aggressively to pre-empt a full-scale epidemic. High STD prevalence in Madagascar makes it vulnerable to the rapid spread of HIV (box 2.9). The epidemic in most of North Africa and the Middle East is also nascent, although there is evidence of rapidly climbing HIV infection among injecting drug users in Bahrain and Egypt, as has been the case in Asia. A large number of North African and Middle Eastern countries could not be classified because of lack of data.

Latin America and the Caribbean

More than half of the countries in Latin America and the Caribbean have concentrated epidemics (figure 2.14). These include the most populous countries in the region—Brazil and Mexico. Six countries have nascent epidemics, two (Guyana and Haiti) have generalized epidemics, and two (Bolivia and Panama) have insufficient information to be classified.

Injecting drug use and sex between men have played a major role in transmission in many countries in Latin America. Roughly one-quarter of all HIV infections in Brazil (24 percent, 1992) and a third in Argentina (39 percent, 1991) have been attributed to transmission through injecting drug use, which is an important source of transmission in Uruguay as well (Bastos 1995, Libonatti and others 1993). The epidemic is well established among homosexual and bisexual men in Argentina, Brazil, Colombia, Mexico, and Peru and has infected significant numbers of sex workers in Argentina, Brazil, the Dominican Republic, Guyana, Honduras, Jamaica, and Trinidad and Tobago. Although the data are spotty, the relatively high prevalence of HIV among injecting drug users, homosexual and bisexual men, and sex workers in Latin America suggests that in many of these countries the virus is poised to spread to the low-risk sexual partners of people who engage in high-risk behavior.

In the Caribbean and parts of Central America, HIV is spread mostly through heterosexual transmission. Male and female cases are roughly equal in Haiti; the epidemic has spread broadly to 8 percent of pregnant women, and there is significant mother-to-child transmission. More than 70 percent of AIDS cases in the Dominican Republic is attributed to heterosexual transmission; the ratio of male-to-female cases now stands at 2 to 1 and is declining (ONUSIDA 1997). HIV prevalence among pregnant women in that country has risen to a national average of 2.8 percent, and in some areas has reached 8 percent. Following a similar path, 1 percent of pregnant women in Honduras are also infected with HIV. In Guyana, which is in South America but faces on the Caribbean, nearly 7 percent of women attending antenatal clinics were infected, as of 1992.

Asia

In most Asian countries for which there is information, the epidemic has reached a concentrated stage either nationwide or at least in some states or provinces (figure 2.15). This includes regions of the world’s two most populous countries, China and India, most of Indochina, and Malaysia. In the remaining Asian countries for which there is information, the epidemic is nascent; infection among those presumed to practice high-risk behavior is less than 5 percent. 

Patterns of infection in east, south, and southeast Asia have been greatly influenced by the proximity of many countries to the “Golden Triangle” of heroin production, located at the border between Lao PDR, Myanmar, and Thailand, and to its distribution routes (background paper, Riehman 1996). HIV infection was first detected among those who inject drugs in Bangkok in 1987; during the next year it spread rapidly among injecting drug users in the Thai capital (Stimson 1994). The pattern was quickly repeated among injecting drug users in northern Thailand and along the border areas between southern Thailand and northern Malaysia. In 1989, HIV infection was identified in Myanmar, Yunnan Province in China, and in Manipur State in India. HIV was detected among injecting drug users in Singapore in 1990.

Injecting drug use has been the main transmission mode in China, where the most highly infected province, Yunnan, is adjacent to international drug routes. Male injecting drug users in Yunnan account for 78 percent of HIV infections in China (Zheng 1996). In other Chinese provinces, infection rates are thought to be low, even among those who practice high-risk behavior (Yu and others 1996). Economic reforms that have helped to reduce the number of people in poverty in China by more than half since the late 1970s have also resulted in large increases in internal migration that could generate conditions conducive to the spread of HIV. Studies have estimated that nearly 100 million people, roughly one in twelve people in China, have moved either temporarily or permanently from their registered residence (Nolan 1993, Peng 1994). Much of the movement involves migration within provinces, but an estimated 20 million migrants have moved from poor areas of western China to eastern provinces (Nolan 1993). Most migrants are young, single, and male, but many women have also migrated; some have reportedly become involved in prostitution. STDs, which were all but eliminated in China in the 1960s, are rising rapidly (Cohen and others 1996, Kang 1995). Early preventive interventions for migrants and sex workers in areas receiving migrants could reduce the likelihood of an epidemic of HIV and other STDs among these mobile groups.

Among the nations of South Asia, the epidemic is believed to be spreading most rapidly in India and Pakistan. In India, HIV is widespread among injecting drug users in the northeastern states of Manipur and Mizoram and is spreading to their sexual partners; prevalence in antenatal clinics in Manipur has reached 2 percent. HIV is well established among sex workers and STD patients in much of southern India, including populous Maharashtra and Tamil Nadu states (Jain, John, and Keusch 1994). In the city of Mumbai (formerly Bombay), HIV prevalence among pregnant women has reached 1 to 2 percent. In Pakistan, the infection rate among injecting drug users in Lahore was 12 percent; as of 1995, HIV infection among women attending antenatal clinics was still extremely low. Transmission by those who inject drugs also may be a factor near a second major heroin-producing area, the “Golden Crescent,” where Pakistan’s Northwest Frontier meets the Badakhshan area of Afghanistan and the Baluchistan area of Iran (background paper, Riehman 1996). However, there are no recent data on HIV prevalence among drug users or other groups in these areas. In Nepal, prevalence so far remains very low among injecting drug users in Katmandu, partly because of interventions discussed in the next chapter. Bangladesh’s HIV epidemic is still nascent, but, without behavior change, HIV could spread quickly among a population of brothel-based sex workers and their clients.

In most of southeast Asia, with the significant exceptions of Indonesia, Lao PDR, the Philippines, and Papua New Guinea, the HIV epidemic is at the concentrated stage. Injecting drug use has played a central role in the launching of HIV, often in conjunction with commercial sex, but heterosexual transmission is now the predominant mode of transmission. HIV is firmly established among injecting drug users and sex workers in Cambodia, Myanmar, and Thailand, and 1 to 3 percent of pregnant women are HIV-positive in those countries. In Thailand, HIV prevalence peaked at 4 percent among military conscripts in 1993, but has recently been declining following a national campaign to reduce sexual transmission of HIV through greater condom use and a reduction in commercial sex. In Cambodia, however, infection levels in the military have reached nearly 7 percent. In Malaysia and Vietnam, more than three-quarters of HIV infections are attributed to transmission through injecting drug use (Hien 1995, Kin 1995). Yet sexual transmission in Malaysia is clearly on the rise; nearly 40 percent of HIV/AIDS cases seen at the University of Malaya Medical Center since 1986 were thought to be due to heterosexual transmission (Ismail 1996). In contrast, although HIV has been detected sporadically for some time among sex workers in the Philippines and Indonesia, it has not spread rapidly, even within that group; as of mid-1996 these two populous countries remained at the nascent stage (Jalal and others 1994, Tan and Dayrit 1994).

Eastern Europe and the Former Soviet Union

The rapid social change and economic dislocation that has accompanied the collapse of socialism in Eastern Europe and the former Soviet Union (FSU) have created a situation in which the potential for an HIV epidemic looms large. The available data on HIV prevalence suggest that most countries in the region are still in the nascent stage (figure 2.16). However, reliable information on HIV prevalence by subpopulation is scarce in all but a handful of countries—almost two-thirds of all of 
the countries in this region cannot be classified based on the available information.

Ukraine has a concentrated HIV/AIDS epidemic, based on high HIV prevalence among injecting drug users; between January and August 1995, HIV prevalence among those who inject drugs rose to 13.0 percent, from just 1.4 percent. Just five months later, more than half of the injecting drug users in the Ukrainian city of Nikolayev were infected (UNAIDS 1996d). A survey of new injecting users in Poland in 1995 found HIV prevalence of 4.7 percent; a few years earlier, among longer-term injectors in the city of Warsaw the rate was 45 percent (WHO/EC Collaborating Centre 1996b). Given that those who inject drugs often travel to neighboring countries, it is reasonable to expect a similarly rapid take-off among injecting drug users in the Russian Federation and Belarus (Bourdeaux 1996).

In Romania, HIV initially spread primarily among children; over 90 percent of AIDS cases in 1990 were among children under 13 years of age. It was erroneously believed that transfusing blood among children would provide important nutrients and boost the immune system (Hersh and others 1991). Instead, it spread HIV among them. The practice has since been abandoned.

A key signal of the potential for an HIV epidemic in this region is the dramatic increase in STDs experienced by most countries since the collapse of the Soviet Union. The number of gonorrhea cases has nearly doubled between 1990 and 1994 in four Eastern European countries (figure 2.17). In Ukraine, the number of syphilis cases increased more than tenfold between 1991 and 1995 (AIDSCAP and others 1996).

Regardless of the stage of the epidemic, there are compelling reasons for governments to find ways to encourage people who practice risky behavior to adopt safer practices as soon as possible. Fortunately, in most areas of the developing world it is not too late to avert a generalized epidemic. Half of the population of developing countries—2.3 billion people—live in areas where the HIV/AIDS epidemic is still nascent. Another third of the population of developing countries lives in areas where the epidemic is already concentrated but has yet to become generalized. In all these areas, action now to help people with the highest-risk behavior protect themselves and others from HIV infection can save millions of lives and avoid massive future expenditures on AIDS treatment and care. Even in areas where the epidemic is already generalized, action to prevent infection among those most likely to contract and spread HIV can still make a substantial difference.

What steps can be taken to help people who engage in high-risk sexual activity or injecting drug use to protect themselves and others from HIV infection? The next chapter discusses two broad complementary approaches: altering the perceived costs and benefits of individual choices, and changing the social environment that shapes and constrains these choices.
 
Previous: AIDS: A Challenge for Government
Next: Efficient and Equitable Strategies for Preventing HIV/AIDS

¹⁷ We recognize that the characteristics of certain subpopulations, like military service or sexual orientation, are imperfect predictors of risky behavior. While members of these subpopulations may practice riskier behavior on average, in some countries they practice low-risk behavior, as evidenced by lower rates of partner change, high rates of condom use, or limited sharing of injecting equipment.

¹⁸ An exception would be adolescents, who can be presumed to be uninfected as they enter adulthood. HIV prevalence among adolescents at a point in time is likely to reflect recent infection.