Not all infectious agents introduced into a population will be self-sustaining. If each infected person transmits the infection, on average, to less than one other person over his or her lifetime, then the infection will eventually disappear; if to more than one other person, then the infection will expand. The reproductive rate of a sexually transmitted disease is the average number of susceptible people infected by an infected person over his or her lifetime (May and Anderson 1987, Thomas and Tucker 1996).¹ If each person infected with a disease transmits it to exactly one other person, then the reproductive rate is 1. In populations in which HIV has a reproductive rate of less than 1, the epidemic will not be self-sustaining. Thus, the greater the reproductive rate of HIV, the more rapidly the epidemic will spread.

What factors, then, determine the HIV reproductive rate in various populations? We have seen in chapter 1 that the most common mode of transmission of HIV is through sexual contact. Three main factors have a large influence on the reproductive rate of all sexually transmitted diseases (STDs), including HIV:

• the amount of time a person remains infectious
• the risk of transmission per sexual contact
• the rate of acquisition of new partners.²

These factors are similar for transmission through contaminated injecting equipment, except that the risk of transmission per contact refers to the risk per injection, and the number of partners refers to number of people with whom injecting equipment is shared. The broad points in the following discussion therefore apply to transmission through sharing of contaminated needles, as well as through sexual contact.

Each of these three factors is in turn influenced by the biology of the virus and by individual behavior. Biology plays an important role in the amount of time a person remains infectious and in the risk of transmission per contact. But individual behavior also has a strong influence on the risk of transmission per contact—for example, through decisions on condom use, disinfecting shared needles, and seeking treatment for other STDs. And individual behavior has a direct relation with the rate of partner acquisition. Until medical science discovers either a cure or a vaccine, the most important avenue for reducing the spread of HIV will continue to be changing individual behavior.

The Duration of Infectiousness

The lack of a cure and the long duration of infectiousness are the main characteristics that distinguish HIV from most other STDs. The long duration of HIV infectiousness increases the likelihood that an infected individual will pass the infection to others. Further, because a person with HIV typically remains asymptomatic for years, an infected individual and his or her sexual (or injecting) partners are often unaware of the risk of transmission. Thus, the long duration of asymptomatic HIV infection potentially puts many more partners at risk than is the case for other STDs.

The impact of recently developed drugs that extend the lives of people with HIV may lengthen the infectious period. However, if these drugs significantly reduce the viral load, they might reduce the risk of infection per contact. Be that as it may, unless both dramatic medical advances and significant reductions in the costs occur, these new drugs are unlikely to have a significant impact on the duration of infectiousness in developing countries, since few developing countries have the financial or human resources to provide them. That leaves two primary mechanisms for prevention—reductions in the risk of infection per contact and reductions in the acquisition of new partners.

The Risk of Infection per Contact

The average risk of infection with HIV per sexual exposure is much smaller than that for other sexually transmitted diseases; however, because of the long period of infectiousness and numerous cofactors that enhance HIV transmission, the chance that an HIV-positive person who does not take precautions will eventually infect others can be quite high.

The most extensive studies of the risk of HIV transmission per exposure have been conducted in industrial countries. Because of generally superior health levels and the ready availability of treatment for other STDs, the average risk of HIV infection per sexual contact in industrial countries is quite small (table 2.1). For example, the average chance that an infected male will sexually transmit HIV to an uninfected female partner by unprotected vaginal sex is estimated at between 1 and 2 per 1,000 exposures. The risk of transmission from an infected female to an uninfected male partner through unprotected vaginal sex is one-third to one-half as great (Haverkos and Battjes 1992).³ Thus, women are believed to have a somewhat greater probability of becoming infected from an infected male partner than the reverse. Anal sex carries the highest risk, especially for the receptive partner. The risk of transmission in unprotected anal intercourse, based on a study of men, is estimated to be between 5 and 30 per 1,000 exposures for the receptive partner. However, all of these figures very likely underestimate—perhaps severely—the average transmission probability per sexual act. They are generally based on studies of transmission within discordant couples—couples in which one partner is HIV-positive and the other is HIV-negative. Couples that are discordant for a very brief period are not captured in these samples; thus, the most infectious individuals are likely to be excluded. These studies also fail to capture couples in which neither partner tests positive for HIV, but one recently has become infected. Below we review evidence that this may be the period of highest infectiousness. If true, then studies of discordant couples are measuring HIV transmission during a less infectious period (Mastro and de Vincenzi 1996).

A transmission rate per partnership, not taking into account the length of partnership, might be a more realistic measure of the risks of sexual  transmission of HIV within relationships.⁴ A review of studies of per-partner transmission rates among heterosexuals in the United States and Western Europe found an average transmission probability of about 23 percent from men to women, and of about half that rate (12 percent) from women to men (Mastro and de Vincenzi 1996). Yet even these rates are probably lower than those faced in developing countries because many people in developing countries are infected with other STDs that enhance HIV transmission, an issue we discuss below. 

The “per contact” risk of HIV transmission with a commercial or casual partner in developing countries is thus likely to be substantially higher than the figures in table 2.1. Notwithstanding these considerations, the average infectiousness of HIV is believed to be substantially less than that for other STDs. In the case of gonorrhea, for example, the probability that an infected woman will transmit the disease to an uninfected male partner during intercourse is 20 to 30 percent per exposure, while the probability that an infected male will transmit the disease to his female partner is 50 to 70 percent (Hethcote and Yorke 1984). 

Although sexual intercourse is the primary means of HIV transmission in both developed and developing countries, other modes of transmission carry a higher probability of spreading the infection. The chance that a mother will transmit the virus to her infant is variously estimated at 13 to 48 percent. Transmission probabilities through sharing contaminated injecting equipment between infected and uninfected injecting drug users are variable, depending on the specific injecting practices and which equipment is shared. The probability of transmission through an accidental needle stick in a medical setting, when the needle has been exposed to HIV-infected blood, is only about 1 in 250, or 0.3 percent. The transmission rate for transfusion of contaminated blood is nearly 100 percent.

The risk of infection per contact is not a constant; it can be influenced by a variety of factors, some of which may tend to exacerbate the epidemic. We discuss the most important of these below.

Risk may be highest soon after infection. Recent studies suggest that infectivity can vary dramatically according to the stage of HIV infection. The two peaks of infectivity are thought to coincide with the periods of highest viral load—the first and highest within the first few months of infection (before the production of antibodies to the virus) and the second, which is thought to be lower, at the very end of the asymptomatic period, as the body loses its battle with HIV (Pinkerton and Abramson 1996). Studies of homosexual men suggest that an individual faces a 10 to 30 percent chance of becoming infected during a single act of unprotected receptive anal intercourse if his partner is at the early, acute stage of the infection (Jacquez and others 1994).⁵ In the middle stage, the likelihood of infection drops to between 0.01 and 0.1 percent, but at the end stage it rises again to between 0.1 and 1.0 percent. The variation in probability of heterosexual transmission in early and late periods of infection has not been estimated but could have important implications for the size of the epidemic box 2.1). 

Greater risk of transmission immediately after infection may be one reason that the epidemic has taken off so rapidly in some developing countries. In Thailand, the average female-to-male risk of sexual transmission was estimated to be 3 to 6 infections per 100 exposures—much higher than the rates in table 2.1—perhaps because more people in Thailand were in the earliest, most infectious stage of the disease (Mastro and others 1994). In addition, the likelihood of HIV transmission also differs by the type of the virus. HIV-1 is more easily transmitted and has a shorter incubation period than HIV-2 (De Cock and Brun-Vezinet 1996). HIV-1 has many subtypes with specific geographic distributions. However, there is no conclusive epidemiological evidence to date that any of these subtypes are more or less infective than others (Anderson and others 1996, Expert Group 1997).

Untreated STDs raise the risk of HIV infection per sexual exposure. STDs are far more common in developing countries than in industrial countries (table 2.2). Studies in both industrial and developing countries have found that people with current or past STDs are 2 to 9 times more likely to be infected with HIV.⁶ However, because HIV and other STDs are both highly correlated with risky sexual behavior—high rates of partner change in particular—it is difficult to disentangle the extent to which conventional STDs actually enhance the transmission of HIV. 

Nonetheless, there are compelling biological reasons for believing that untreated ulcerative STDs such as herpes, syphilis, and chancroid greatly increase the risk of HIV transmission per exposure: the lesions caused by these diseases provide a ready portal for transmission of HIV, whether they are on the HIV-infected or the uninfected partner. Enhanced HIV transmission in the presence of nonulcerative STDs such as gonorrhea, chlamydia, or trichomoniasis is also biologically plausible but the epidemiological evidence to support it has been weaker, mainly for methodological reasons (Laga and others 1993). For example, a recent study in Malawi found that the amount of HIV virus in the semen of HIV-positive men with urethritis was eight times higher than in a control group of HIV-positive men without it, and these concentrations diminished significantly when the urethritis was treated with antibiotics (Cohen and others 1997). Women are more likely than men to have STD infections without any apparent symptoms, and therefore many infections go untreated. Half of all women with gonorrhea, for example, have no symptoms, compared with only 5 percent of men (Hethcote and Yorke 1984). Thus, if nonulcerative STDs do facilitate HIV transmission, they are likely to differentially raise the transmission probabilities to and from women, since a higher percentage of women are likely to have asymptomatic STD infections that go untreated. A recent review found that, in eleven African countries, from 5 to 17 percent of pregnant women tested positive for syphilis; in Jamaica the rate was 5 percent, and in Haiti more than 10 percent (Van Dam, Dallabetta, and Piot 1997).

Whatever the precise nature of the link between HIV and other STDs, there is evidence that treating symptomatic STDs reduces HIV transmission. In the early 1990s a randomized controlled trial in rural areas of Mwanza region, Tanzania, found that treatment of symptomatic classic STDs lowered the incidence of HIV among adults by more than 40 percent (Grosskurth and others 1995a). The extent to which this result can be generalized to other countries is likely to depend on many country-specific factors, including the underlying prevalence of HIV and STDs, the types of STDs that are prevalent, the quality of treatment services, and existing levels of STD treatment prior to the intervention. At the outset of the Mwanza study, HIV prevalence among adults 15 to 54 was already high—4 percent (Grosskurth and others 1995b). Simulations of the HIV epidemic in rural Uganda indicate that the proportion of HIV infections for which STDs were a cofactor was highest early in that epidemic (Robinson and others 1997). This suggests that the effectiveness of STD treatment in slowing HIV incidence in Mwanza might have been even greater had it occurred much earlier in the epidemic.

Male circumcision may be a factor. Some researchers have found a correlation between HIV infection and lack of circumcision among men and believe that this may account in part for the rapid spread of HIV in Sub-Saharan Africa.⁷ Ethnographic studies suggest that men are least likely to be circumcised in central, eastern, and southern Africa, along a north–south swath through the Rift Valley (Bongaarts and others 1989). This also happens to be the area with the highest rates of HIV infection in urban areas. In 1989, in five countries where more than three-quarters of men were not circumcised, the urban prevalence of HIV was roughly 16 percent. In contrast, the average level of urban HIV infection was only 1 percent in 20 other countries where more than 90 percent of men were estimated to be circumcised.

One reason why uncircumcised men could be at higher risk of contracting HIV and passing it to others is that they are at higher risk of developing ulcerative STDs, particularly chancroid. Poorer genital hygiene among uncircumcised men may also play a role, particularly in low-income and unsanitary settings. A study in Kenya found that, even among men without chancroid, uncircumcised men were more likely to sero-convert (29 percent) than those who were circumcised (2.5 percent) (Plummer and others 1991). However, the amount of increased risk of HIV infection from lack of circumcision alone has not been established and whether or not such a risk exists is still debated. This is because circumcision is highly correlated with many other factors besides chancroid. In particular, ethnicity and religion are strong determinants of whether or not men are circumcised. It is therefore difficult to disentangle the effect of male circumcision (or lack of it) from that of other cultural norms that affect sexual behavior.

Clearly, even if male circumcision is protective against acquiring and spreading HIV, it is not sufficient to prevent infection. High proportions of men in West Africa are circumcised, yet HIV has nonetheless spread rapidly there. More than three-quarters of U.S.-born men are circumcised, but that has not prevented a sexually transmitted HIV epidemic in the United States (Laumann, Masi, and Zuckerman 1997). Conversely, in Western Europe and South America, circumcision is uncommon, yet the HIV epidemics in those areas have not reached the scale of the one in eastern and central Africa (de Vincenzi and Mertens 1994). 

Behavior affects the probability of transmission. Although the basic transmission probabilities for HIV per exposure are founded in the biology of the virus, fortunately they can be substantially reduced through behavioral change. Using latex condoms and obtaining treatment for conventional STDs can lower the probability of transmission through sexual contact. Sterilization of injection equipment can dramatically reduce transmission among injecting drug users and among patients in medical facilities. And mother-to-child transmission can be reduced through both medical treatment and behavioral changes. The prospects for changing behavior to reduce HIV transmission are discussed in chapter 3.

The Rate of Partner Change

While transmission probabilities have an important influence on the reproductive rate of HIV, the rate of sexual partner change probably accounts for the greatest differences in the rate across groups and countries. Similarly, the rate at which injecting drug users change partners with whom they share unsterilized injecting equipment strongly influences the HIV reproductive rate among them. Finally, in medical settings, the rate of reuse of unsterilized injecting equipment for multiple patients is analytically equivalent to the rate of partner change (box 2.2). In all three situations, the higher the rate of partner change is, the greater is the likelihood that the virus will pass from infected to uninfected people. 

In contrast, although the probability of becoming infected per exposure is higher for people receiving transfusions of infected blood and for the children of HIV-positive mothers, these groups are unlikely to infect many others. The rate of partner change among transfusion recipients, for example, is quite low, on average. Because the HIV reproductive rate for these modes of transmission is probably less than 1, if the virus were spread only by transfusion or from mother to child, the epidemic would most likely not be sustained. In the absence of condom use or sterilization of shared injecting equipment, rapid rates of partner change sustain the epidemic.

Both the average rate of partner change in a population and the variation of the rate across individuals have an impact on the spread of HIV in populations. Other factors being constant, the higher the average rate of partner change is, the higher is the reproductive rate of HIV. However, in a population in which a few people have very high rates of partner change and many people have very low rates, HIV and other STDs will spread more quickly than if the same average number of partners were distributed more equally across the entire population (Anderson and May 1988, Over and Piot 1993).

Surveys of sexual behavior suggest that there is in fact quite significant variation in the rate of partner change across subgroups in a single population.⁸ To take one example, figure 2.3a, and figure 2.3b shows the distribution of men and women 15 to 49 in Rio de Janeiro, Brazil, who had at least one nonregular sexual partner in the previous twelve months, according to the total number of nonregular partners. There are two “peaks” in the distribution of people by their number of nonregular partners—one large peak among those with no nonregular partners or only a few, and another small peak among those with very many partners. Roughly half of men (56 percent) and 90 percent of women reported no nonregular partner, meaning that they either did not have any sexual partner or that they had sex only with their spouse or with some other regular partner. Those who did have nonregular partners usually acknowledged only a few. For example, roughly 12 percent of men and 6 percent of women reported only one nonregular partner in the past twelve months. On the other hand, a small percentage of men—nearly 2 percent—reported having 20 or more partners in the same period. The “two-peaked” distribution of people according to their nonregular sexual partners is typical of those found from sexual behavior surveys in other countries (background paper, Deheneffe, Caraël, and Noumbissi 1996).

The variation, or heterogeneity, in sexual behavior is even more striking in Thailand, where, in 1990, 28 percent of men 15 to 49 had a nonregular partner in the past year and almost 4 percent had 20 or more partners, while only 2 percent of women reported any nonregular partner.⁹ Sampling methods for these surveys are typically less successful in capturing sex workers, who are usually a small percentage of the population but an important component in the second peak of the bimodal distribution. High rates of partner change in a very small subgroup may be sufficient to sustain an STD or HIV epidemic that can gradually spread to the rest of the population.

Mixing patterns. The path of the epidemic within the overall population depends on the degree and pattern of mixing among people with high-risk behavior, and the mixing between people with high-risk behavior and people with low-risk behavior. By “high-risk behavior,” we mean unprotected sexual intercourse with multiple partners or sharing of unsterilized injecting equipment. People with high-risk behavior are very likely to become infected and to unknowingly pass HIV to others. People with low-risk behavior—who have few partners, who consistently use condoms, who do not inject drugs, or (if they do) do not share injecting equipment—are less likely to pass HIV to others. However, they are nevertheless at risk of becoming infected through transfusion of contaminated blood or sexual mixing with people who practice high-risk behavior. And young children are at risk of becoming infected from their mothers perinatally but are highly unlikely to spread HIV.

In a sexually-transmitted HIV epidemic, the speed at which HIV spreads from people with a large number of partners to those with very few partners depends on the extent of mixing between people with different levels of sexual activity. If people with large numbers of partners have intercourse only with others who are similarly active (known as assortative sexual mixing), then HIV will tend to rise rapidly within those groups but only very slowly and to a limited extent in the rest of the population. As a result, the epidemic will achieve lower peak levels of infection in the entire population than if those with large numbers of partners also have sex with those who have fewer partners (known as random or disassortative mixing) (Anderson 1996; Anderson, Gupta, and Ng 1990). Mixing patterns explain why HIV does not spread through a population at a uniform rate. Rather, it spreads in a series of smaller epidemics that race through overlapping subpopulations whose behavior puts them at various degrees of risk, then outward to those with less-risky behavior with whom they mix.

Sex workers whose clients do not use condoms, injecting drug users who share unsterilized injecting equipment, and others with very high rates of partner change are typically the first to be infected in an HIV epidemic. HIV prevalence in these groups can rise very rapidly. Figure 2.4 shows rapid increases in HIV prevalence among sex workers in several cities in developing countries. Some of the differences in the rate of increase across cities can be attributed to differences in the timing of the introduction of the virus. However, other factors also appear to be at work. HIV prevalence among sex workers in Santo Domingo, Dominican Republic, has risen more slowly; this is believed to be due to very high rates of condom use among brothel-based sex workers (Peggy McEvoy, personal communication). 

HIV tends to move even more rapidly among injecting drug users who share injecting equipment than among sex workers because the risk of transmission per contact is much greater. In countries where those who inject drugs commonly share injecting equipment, HIV can infect the majority of users in a matter of months, as has been the case in parts of Asia and in Ukraine (figure 2.5). 

The following figures, ( Figure 2-6a, Figure 2-6b, Figure 2-6c, Figure 2-6d, Figure 2-6e, and Figure 2-6f ), shows the incidence of AIDS in various population groups in six areas of Latin America. In Brazil, the first wave of the epidemic was among men who have sex with men.¹⁰ This was followed a few years later by an epidemic among injecting drug users, most of whom are also men. Later, the disease spread to sex workers and the female partners of bisexual men and injecting drug users.

The timing and pattern of the epidemic waves can be very different, even within a single region. In the Andean Area, Mexico, and the Southern Cone, the epidemic first struck men who have sex with men, as it did in Brazil. In the Caribbean and the Central American Isthmus, heterosexual transmission quickly outpaced transmission in other groups. In Thailand, HIV spread first among men who have sex with men and injecting drug users, and then spread among sex workers and their clients. Researchers have since determined that the epidemics among injecting drug users and sex workers in Thailand were largely independent, spreading two distinct variants of the virus (Ou and others 1993).

Subpopulations that practice high-risk behavior are constantly in flux (Weniger and Berkley 1996). Over time, some individuals abandon high-risk behaviors or die, while others initiate high-risk behavior, adding to the high-risk subpopulation. High-risk behavior usually varies over an individual’s life cycle. Sexual activity is often highest among young, unmarried adults, particularly men. As men and women marry and age, levels of casual sex typically decline. The age profile of the percentage of men and women engaging in sex with a nonregular partner in figure 2.7 demonstrates this well, although it also reflects to some extent temporal changes in social norms. As we shall see in chapter 3, changing socioeconomic factors can also induce people to adopt or abandon risky behavior. The dynamic nature of the subpopulation practicing high-risk behavior at any point in time prevents their HIV-infection rate from reaching 100 percent.

If there is very little mixing between people with different degrees of risky behavior, the overall epidemic may have multiple peaks. Incidence may rise and fall several times as first one group and then another becomes nearly saturated by the virus. Thus, a sustained decline in incidence in a specific group of people with risky behavior does not necessarily signal the end of the epidemic in the entire population (Anderson 1996; Anderson, Gupta, and Ng 1990).

Concurrent partnerships. Partnerships that overlap in time are concurrent partnerships. Examples of concurrent partnerships include: partnerships between married men or women and commercial or casual sexual partners; people engaging in long-term relationships with more than one casual partner; and polygyny, the practice of having more than one wife. In two populations in which individuals have the same average number of partners in a given period, HIV and other STDs will spread more rapidly in the population in which partnerships are concurrent than in the population in which partnerships occur sequentially (background paper, Morris 1996). This is because when partnerships are sequential, the virus cannot spread to a new susceptible person until the dissolution of one relationship and the start of another. In a concurrent partnership, it can infect more susceptible people in a shorter period of time.

Recent research has focused on the role of “bridge populations” in the spread of HIV. These are partnerships that link people in groups that otherwise might not mix, such as partnerships between people with high-risk behavior and those with very low-risk behavior (background paper, Morris 1996; Morris and others 1996). For example, men who have unprotected intercourse with sex workers and have either a wife or steady girlfriend may transmit HIV to monogamous women who would not otherwise be at risk. The extent of such mixing in Thailand was recently captured by a survey of sexual behavior among low-income men and long-haul truckers in three provinces (table 2.3). The bridge population comprises a very large percentage of both groups—about 17 percent of low-income men and 25 percent of the truckers. Sexually active injecting drug users are another potential bridge population. In Manipur state, India, within two years of the first reported HIV case among drug injectors, 6 percent of the noninjecting sexual partners of injecting drug users were infected (Sarkar and others 1993).

The impact of heterogeneity, mixing, and concurrency: A simulation. The combined effect of heterogeneity in sexual behavior, mixing between groups of different behaviors, and concurrent partnerships can have a profound impact on the potential course of the HIV/AIDS epidemic, absent any intervention. Figure 2.8 shows simulations of heterosexual HIV epidemics in four imaginary populations with different underlying patterns of sexual behavior (background paper, Van Vliet and others 1997). The four simulated populations are identical in every respect, except in their patterns of sexual behavior, as follows:

• Commercial and casual sex: In this population, some men engage in commercial and casual sexual relationships before and after marriage; some women have relations with casual partners before marriage but all women are monogamous after marriage. The commercial and casual partnerships can be concurrent with each other and with marriage. Sex workers comprise about a one-fifth of 1 percent of all women.

• Commercial sex: Some men engage in commercial sex before and after marriage with a small group of female sex workers comprising only one-quarter of 1 percent of the female population. Neither men nor women have casual sexual relationships before or after marriage. Commercial and marital relationships can be concurrent.

• Casual sex: Some men and women have casual sexual relationships before and after marriage. There are no sex workers in this population, but casual partnerships can be concurrent with each other and with marriage.

• Serial monogamy: Some men have commercial and casual sexual relationships before marriage and some women have casual relationships before marriage. After marriage, neither men nor women have extramarital relationships. All relationships before marriage are serial; this is the only one of the four populations with no concurrent relationships. Sex workers are 0.05 percent of the female population.

Using the STDSIM simulation model described in box 2.3, it is possible to see how the course of the epidemic would differ in these four imaginary populations with no specific intervention or behavior change.¹¹ Baseline levels of condom use are assumed to be 5 percent among casual partners and 20 percent among sex workers. HIV is introduced in the population in year zero.

The results in figure 2.8 show the trends in HIV prevalence for the entire population, including members with high- and low-risk behavior. The first curve, at the top of the figure, shows the path of the epidemic in the population in which people have concurrent relationships and commercial and casual sex. Thirty years into the epidemic, HIV begins to show signs of leveling off, but at a very high level—30 percent. In the population in which there is only commercial sex and marital sex, HIV prevalence peaks eight years after the virus is introduced at about 13 percent, then declines to an equilibrium prevalence of about 8 percent 20 years into the epidemic. Prevalence declines even in the absence of any behavior change because the people with the riskiest behavior, infected at the outset, begin to die. Most new cases of HIV infection at that point occur among people who have recently adopted high-risk behavior. In the population in which there is casual but no commercial sex, the epidemic progresses more slowly, reaching a prevalence rate of only about 3 percent eight years after the start of the epidemic. However, prevalence continues to climb, reaches 30 percent, and is still increasing 30 years after HIV is introduced. Finally, in the population that practices serial monogamy with commercial, casual, and marital partners but no concurrent partnerships (“serial monogamy,” the lowest curve), HIV prevalence rises more quickly than in the population with only casual sex, but levels off at about 9 percent. Thus, even without behavior change to prevent HIV, the course of an HIV epidemic can be quite different across populations, depending on the heterogeneity of behavior, the extent of mixing, and the degree of concurrency in partnerships.

Previous: HIV Incidence and Prevalence and AIDS Mortality
Next: Implications for Public Policy
 

¹ Thomas and Tucker (1996) point out that the use of the word “rate” is not technically correct, since it is in fact a number, not a rate per unit of time. The reproductive rate was first applied to HIV transmission by May and Anderson (1987). Previous authors had referred to this as the “infectee number” (Hethcote 1976, Nold 1978).

² The number of exposures per partner and the type of sexual act also affect the spread of HIV through the population, but empirical evidence suggests that the rate of partner change is far more important (Anderson, Gupta, and Ng 1990).

³ However, a review of sixteen studies that compared female-to-male and male-to-female transmission found female-to-male rates as low as 5 percent and male-to-female transmission rates as high as 140 percent (Haverkos and Battjes 1992). 

⁴ Transmission rates per partnership are not much affected by the length of the partnership or number of exposures. This is because in a relatively long-term partnership, either the HIV-positive partner is truly infectious and transmission will occur relatively quickly, or the HIV-positive partner is relatively uninfectious and transmission will not occur despite many contacts over a long period.

⁵ High infection rates with other STDs may also have contributed to high transmission rates in the study.

⁶ For example, Hook and others 1992, Laga and others 1993, Lazzarin and others 1991, Mastro and others 1994, Plummer and others 1991, Quinn and others 1990.

⁷ For example, Bongaarts and others 1989; Caldwell and Caldwell 1996; Conant 1995; de Vincenzi and Mertens 1994; Moses and others 1990, 1995; Simonsen and others 1988.

⁸ Like all data on sexual activity, such surveys are subject to a variety of errors. Since they ask about private behavior, misreporting may skew results. In many cultures men may exaggerate the number of their sexual partners, while women may do the opposite. People with high rates of partner change may not be able to recall their number of partners accurately, and sex workers may not be captured by surveys using standard sampling techniques. Even if we assume a significant degree of error, however, the variation in the rate of partner change within populations is striking.

⁹ By 1993, the percent of men 15–49 with nonregular sexual partners in the past twelve months had declined to 15 percent in Thailand (Thongthai and Guest 1995).

¹⁰ Men who have sex with men include self-identified homosexual and bisexual men as well as heterosexual men who have sexual relations with other men.

¹¹ The impact of behavior change on the epidemic will be introduced later in this chapter and in chapter 3.