Action under Uncertainty
Even though the current deficit in evaluation research is glaring, the magnitude and seriousness of the global pandemic means that action is nevertheless required. Moreover, despite such gaps in knowledge, we can still improve control strategies by tailoring interventions to the nature and scope of the epidemic. Summarized below is what is known with regard to the burden of disease, the determinants of transmission, and the effectiveness and cost-effectiveness of existing prevention interventions.
Burden of Disease
As a result of large-scale implementation of data collection methods for surveillance worldwide and enhanced methods for validating and interpreting HIV-related data, the HIV/AIDS epidemic is probably one of the best documented epidemics in history. An increasing number of data sources contribute to reasonably accurate estimates and a more nuanced understanding of the epidemic's trends. Unfortunately, this relatively accurate picture of where the epidemic is and has been is not matched by similarly convincing maps of the factors that explain its spread.
Although no single country has been spared the virus, the epidemic has affected certain regions of the world disproportionately, and Sub-Saharan Africa remains by far the hardest hit region (table 18.1). With only 10 percent of the world's population, it accounts for more than 75 percent of all HIV infections worldwide and more than 75 percent of AIDS-related deaths estimated for 2003. Asia and the Pacific, with several large and populous countries, account for 7.4 million infections, or 19.5 percent of the current burden of disease. Prevention and treatment efforts in Sub-Saharan Africa and Asia—regions that together represent 85 percent of all current infections—have dictated, and will continue to dictate, global trends in the burden of HIV- and AIDS-related mortality.
Between 1997 and 2001, the percentage of women living with HIV/AIDS increased from 41 to 50 percent. This trend is most apparent in Sub-Saharan Africa, where women represent 57 percent of adults living with HIV and 75 percent of HIV-infected young people. Even though women account for a smaller share of infections in Asia (28 percent), the disease burden among women and girls is likely to rise as the epidemic becomes generalized. More detailed information about the global burden of HIV/AIDS, regional differences, and trends over time is available in the UNAIDS (2005) report on the global AIDS epidemic.
Determinants of Infection
HIV transmission predominantly occurs through three mechanisms: sexual transmission, exposure to infected blood or blood products, or perinatal transmission (including breast-feeding). The likelihood of transmission is heavily affected by social, cultural, and environmental factors that often differ markedly between and within regions and countries. There is also some indication that molecular, viral, immunological, or other host factors might influence the likelihood of HIV transmission. For a more detailed discussion of sexual behaviors and the contextual determinants of infection, see chapter 17.
Worldwide, sexual intercourse is the predominant mode of transmission, accounting for approximately 80 percent of infections (Askew and Berer 2003). Sexual intercourse accounts for more than 90 percent of infections in Sub-Saharan Africa. Although many people who know they are infected reduce their risk behaviors, studies in developed countries suggest that a substantial percentage nevertheless continue to engage in unprotected sex (Marks, Burris, and Peterman 1999). The risk of sexual transmission is determined by behaviors that influence the likelihood of exposure to an infected individual and by infectivity in the event of exposure. This also includes factors related to the infectiousness of the infected partner and the susceptibility of the uninfected partner.
The per contact infectivity of HIV from sexual transmission varies depending on sexual activity (Royce and others 1997). Anal intercourse carries a higher transmission probability than penile-vaginal intercourse, and male-to-female transmission is more likely than female-to-male transmission. Data on infectivity by transmission mode are shown in table 18.2.
Biological Mediators of Infectivity
Untreated STIs increase the risk of sexual HIV transmission several-fold (Institute of Medicine 1997). Numerous epidemiological studies have supported the association of genital ulcers in general and of genital herpes (herpes simplex virus 2, or HSV-2) in particular with HIV infection (Hook and others 1992). Not only does the biological interaction between HSV-2 and HIV enhance the transmission and acquisition of HIV, but HIV infection is also associated with more frequent reactivation of HSV-2. The presence of herpetic ulcers and lesions allows an entry point for HIV in the uninfected individual, and the presence of high copy numbers of HIV ribonucleic acid (RNA) in HSV-2 lesions in HIV-infected individuals underscores the importance for HIV prevention of controlling HSV-2 infections (Mbopi Keou and others 1999).
Vaginal infections are also emerging as important risk factors for HIV. For example, infection with trichomonas increases the risk for HIV seroconversion (Buve 2002). In addition, higher trichomonas rates have been detected in regions of Sub-Saharan Africa that have higher HIV rates, and investigators working throughout Sub-Saharan Africa report similar results, with odds ratios from 1.5 to 56.8 (Gregson and others 2001). In addition, studies have shown an increased risk of HIV acquisition in patients who have bacterial vaginosis (Martin and others 1999).
Circumcision also affects HIV transmission. In a meta-analysis of 27 studies (Weiss, Quigley, and Hayes 2000), uncircumcised men were almost twice as likely to be infected with HIV as those who were circumcised. Studies that controlled adequately for other risks and studies that separately assessed risk in high-risk populations, such as STI clinic attendees or truck drivers, found an even stronger protective effect of circumcision. Similarly, an ecological study comparing two high-prevalence Sub-Saharan African cities with two low-prevalence cities found that circumcised individuals were substantially less likely to be infected with HIV (Auvert and others 2001). Two recent studies conducted in Kenya and India (Donnelly 2004; Reynolds and others 2004) found that uncircumcised men had an HIV rate 7 to 11 times greater than circumcised men. More recently, results from a randomized controlled trial conducted in South Africa indicated that the risk of HIV acquisition was reduced by more than 60 percent of men randomized for circumcision (controlling for sexual behavior, including condom use and health seeking behavior) in a community where more than 30 percent of the women were infected (Auvert and others 2005).
Before circumcision among adult males becomes a widespread policy recommendation, results are still pending in two similar trials. Obviously one issue is the acceptability of such a procedure as well as the fact that some increase in high risk sexual activity was noted among the men who were circumcised, although this did not offset the results of the intervention.
The risk of sexual transmission is also strongly correlated with the plasma level of virus in the infected individual (Quinn and others 2000); thus, infectivity varies over the natural progression of the disease. Individuals are most infectious subsequent to infection and again during the late stage of the disease. Antiretroviral therapy significantly reduces the level of virus, often to the point that standard tests cannot detect HIV in the patient's blood (Palella and others 1998). Available data suggest that viral load reductions induced by antiretroviral therapy will lower infectiousness. Studies have shown a close relationship between the amount of viral suppression and the risk of vertical transmission (Garcia and others 1999). Quinn and others (2002) show that the risk of sexual transmission between couples in Africa was strongly related to the level of viral load in the infected partner.
Exposure to Infected Blood or Blood Products
Injection drug use and blood transfusion are two mechanisms of HIV exposure to infected blood. Determinants of each are discussed below.
Because of the efficiency of HIV transmission through needle sharing, the introduction of HIV into an urban network of injecting drugs users can quickly lead to extraordinarily high HIV prevalence in this population. Sharing of injection equipment and frequency of injection are both important correlates of HIV infection (Chaisson and others 1989). Attendance at shooting galleries, where sharing with anonymous injecting partners is likely to occur, is also an independent risk factor across many studies (Vlahov and others 1990). Injecting cocaine (associated with "booting" or "kicking," where blood is drawn into the syringe and then injected) and having a number of needle-sharing partners are also associated with HIV infection (Anthony and others 1991).
The probability of becoming infected through an HIV-contaminated transfusion is estimated at more than 90 percent (UNAIDS 1997), and the amount of HIV in a single contaminated blood transfusion is so large that individuals infected in this manner may rapidly develop AIDS. Currently, between 5 and 10 percent of HIV infections worldwide are transmitted through the transfusion of contaminated blood products (WHO 2002a). Setting up and maintaining a safe blood supply will virtually eliminate HIV transmission through transfusions.
Perinatal HIV transmission includes both vertical transmission and transmission during breastfeeding. Determinants of each are discussed below.
Perhaps the most compelling evidence of the significance of viral load and transmission risk has been documented with respect to MTCT. Maternal viral load, as quantified by RNA polymerase chain reaction, is associated with increased risk in each mode of vertical transmission. A recent randomized clinical trial in Kenya found that maternal plasma HIV RNA levels higher than 43,000 copies per milliliter were associated with a fourfold increase in vertical transmission (John and others 2001).
Independent of HIV RNA levels in maternal plasma, additional risk factors include cervical HIV deoxyribonucleic acid (DNA), vaginal HIV DNA, and cervical or vaginal ulcers. Chorioamnionitis has also been documented as a risk factor for MTCT among African mothers (Ladner and others 1998), as has exposure to maternal blood during labor and delivery. Newell (2003) estimates that for every hour an infant is exposed to ruptured membranes, the risk of transmission increases by 2 percent.
Transmission through breastfeeding is likely associated with an elevated viral load in the breast milk, which in turn is associated with maternal plasma viral load and CD4 T cell levels. Mastitis has also been associated with increased risk of vertical transmission. Meta-analyses suggest that the cumulative probability of HIV infection increases from 0.6 percent at age 6 months to 9.2 percent at age 3 (Read 2003). A study in Malawi, however, indicates that most transmission occurs in the early breastfeeding months, with an incidence per month of 0.7 percent at age 1 to 5 months, 0.6 percent at age 6 to 11 months, and 0.3 percent at age 12 to 17 months (Miotti and others 1999). In one study, infants who were breastfed in combination with receiving other supplementary foods were twice as likely to be infected at age 6 months than infants fed exclusively on breast milk or on formula (Coutsoudis and others 2001). The hypothesis is that antigens and bacterial contaminants present in supplemental fluids and foods consumed by infants who are not exclusively breastfed may cause inflammation and microtrauma to the infant's intestinal gut, thereby facilitating viral transmission. Another hypothesis is that mixed feeding increases the risk of subclinical or clinical mastitis in the mother, which could increase milk viral load (Semba and others 1999).
Decisions about breastfeeding are further complicated by recent data indicating possible increased mortality among breastfeeding mothers (Nduati and others 2001) and by the stigma associated with not breastfeeding in countries where abstaining from breastfeeding is tantamount to disclosing a woman's HIV status.
Effectiveness and Cost-Effectiveness of Prevention Interventions
Below we discuss the need for ongoing surveillance and contextual data to determine the effectiveness of HIV interventions and how best to implement those interventions. We then discuss the existing effectiveness and cost-effectiveness data.
Essential Background Data for Any Intervention
Because the prioritization of prevention strategies for any epidemic requires accurately identifying the epidemiological profile (discussed below), maintaining a sound and reliable public health surveillance system is a prerequisite for an effective prevention response. An understanding of HIV and STI prevalence and trends, as well as the prevalence and distribution of behaviors that contribute to the epidemic's spread, should be supplemented by national monitoring systems that track sources and uses of funding to promote greater accountability. In addition, data are needed to identify and characterize key contextual issues that affect the selection of interventions.
Although surveillance is essential for an optimally strategic public health response, its utility depends on the degree to which the information it yields is effectively deployed. As noted below, countries with concentrated epidemics should prioritize interventions that are targeted to the populations at highest risk. In Latin America, however, where information on national AIDS funding is strongest, the proportion of limited prevention resources that is not targeted to the populations at highest risk of infection varies from less than 5 percent to more than 50 percent (Saavedra 2000). This range strongly suggests that resource allocation is frequently not based on available epidemiological and effectiveness data.
Table 18.3 summarizes information about the effectiveness of the interventions discussed below.
Cost-Effectiveness Estimates for Prevention Interventions
How countries spend funds and which interventions they prioritize should be guided by estimates of the relative cost-effectiveness of such interventions. Unfortunately, reliable estimates of cost-effectiveness are largely lacking, for a number of reasons. The main reason is that HIV prevention interventions are difficult to force into a typology that clearly distinguishes one intervention from another. For example, the counseling component of VCT has a strong information-sharing element that overlaps with (a) information, education, and communication (IEC) through the media; (b) peer interventions; and (c) the counseling component of STI treatment. Similarly, the psychological support offered through counseling is comparable to support provided through support groups or to interventions designed to increase social support. Such overlap and duplication among components of different interventions complicate efforts to estimate both the effectiveness and the cost-effectiveness of different interventions.
Several authors have recently reviewed estimates of cost-effectiveness for the prevention interventions described here (Creese and others 2002; Jha and others 2001; Marseille and others 2002; Walker 2003). These reviews address a number of methodological issues that will not be repeated here. The reviews agree that the availability of cost and cost-effectiveness analyses for HIV/AIDS prevention strategies is limited and that the need for such knowledge for planning and decision-making purposes is urgent.
Table 18.4 summarizes available cost-effectiveness estimates for the four UNAIDS epidemic profiles that are described later in table 18.5. The estimates of cost per disability-adjusted life year (DALY) saved assume a uniform 20 DALYs lost per infected adult (Murray and Lopez 1996) and 25 DALYs lost per infected child (Marseille and others 1999) and do not account for the increasing proportion of people living with HIV/AIDS in developing countries who will have access to antiretroviral therapy over the coming years.
General Interventions Relevant for All Modes of Transmission
The following are general interventions not specifically targeting the mode of transmission:
Information, education, and communication. This intervention includes education on HIV/AIDS and condom use through pamphlets, brochures, and other promotional materials in classroom or clinic settings or through the radio, television, or press. In general, discerning the effectiveness of IEC alone is difficult, because IEC is often included in condom promotion and distribution interventions. Here we consider the effectiveness of IEC in concert with condom promotion and distribution. Of all available prevention interventions, providing information and education about HIV/AIDS is perhaps the most difficult to assess for cost-effectiveness. Numerous studies have shown that information alone is typically insufficient to change risk behavior. Accurate information, however, is indisputably the basis for informed policy discourse—a vital ingredient in the fight against fear-based stigma and discrimination. In the absence of studies to guide the level of investment in IEC, the only reasonable alternative seems to be to implement IEC on the basis of data derived from relative levels of knowledge and understanding in the population. For example, if only 25 percent of the sexually active population were able to describe how HIV is transmitted and prevented, clearly more IEC would be needed, but if 75 percent of the population understood the basic facts about HIV/AIDS, the need for additional funding would be diminished.
School-based sex education. School-based sex education programs, an aspect of IEC, provide information to young people and reinforce healthy norms in a school setting (Peersman and Levy 1998). Limited data have shown differences in students who have been exposed to school-based sex education (summarized in table 18.3). Box 18.2 reviews the effectiveness of abstinence-only education and comprehensive sex education, subsets of school-based sex education. In light of more recent controlled studies that have not shown an effect on condom use, STIs, or HIV infection, any cost-effectiveness estimate is extremely speculative.
Voluntary counseling and testing. This intervention enables people to know their HIV status and provides counseling support to help them cope with the outcome. Knowledge of serostatus may lead individuals to avoid engaging in risky behaviors (Sweat and others 2000). Cost-effectiveness estimates of VCT vary widely, and as with many other prevention interventions, these estimates are extremely sensitive to the prevalence of HIV in the population that is seeking testing.
Peer-based programs. Peer interventions use influential members of a targeted community to disseminate information or teach specific skills. Such interventions have generally been found to be effective in reducing unsafe behaviors. Work on the cost-effectiveness of peer-based interventions in developing countries has been minimal. In Chad, Hutton, Wyss, and N'Diekhor (2003) reviewed data on 12 prevention interventions and integrated them into a comparative analysis. Their findings suggest that peer education for sex workers is likely to be highly cost-effective and to entail one-fifth the cost of the next most favorable intervention, blood safety. However, the estimated cost-effectiveness for the same intervention directed toward young people and high-risk men is 33- to 36-fold lower.
Interventions to Prevent Sexual Transmission
Below we discuss the effectiveness and cost-effectiveness of interventions that target sexual transmission of HIV:
Condom promotion, distribution, and social marketing. Condom promotion, distribution, and social marketing vary by epidemic profile. The evidence on condom promotion and distribution programs indicates that such programs result in significantly higher condom use and significantly lower STI incidence (see table 18.3). Given the central role that condom promotion, distribution, and social marketing has played in HIV prevention programs, the lack of data on the relative cost-effectiveness of such programs 20 years into their implementation is striking. It is beyond dispute that the use of a condom by sexual partners who are HIV-discordant is extraordinarily cost-effective, given the low cost and high effectiveness of the condom in preventing HIV transmission. Information on the relative costs and effectiveness of different approaches to increasing condom use by serodiscordant sexual partners is not available, with the shortage of information being far more acute for effectiveness than for costs. In the absence of empirical evidence, decision makers are reduced to formulating policy on the basis of theory and common sense. Even inefficient use of condoms by seroconcordant couples is likely to be highly cost-effective because of the reduction in other STIs, cervical cancer, and unwanted pregnancies. However, more reliable information on strategies to optimize the effectiveness and cost-effectiveness of condom programs is urgently needed.
STI screening and treatment. The latest analyses suggest that STI control may be most effective as an HIV prevention strategy when initiated earlier in the course of national epidemics and when sexual risk behaviors are high (Orroth and others 2003). In most developing countries, the greatest benefits from treating STIs almost certainly accrue from averting the morbidity and mortality caused directly by STIs rather than indirectly because of reduced HIV transmission. Estimates of the cost-effectiveness of STI treatment purely as a way to reduce HIV transmission vary widely.
Prevention of Mother-to-Child Transmission
The existing data on the effectiveness and cost-effectiveness of HIV interventions target MTCT in order of decreasing cost-effectiveness as follows:
Avoidance of unwanted pregnancies among infected mothers. One of the most effective strategies to reduce HIV among infants is to provide better contraception services. See box 18.3 for details.
Use of antiretroviral therapy. Evidence indicates that the provision of antiretroviral drugs to infected mothers significantly reduces vertical transmission (see table 18.4). The provision of antiretroviral therapy to prevent MTCT is highly cost-effective, to the point of being cost-saving for women who already know that they are infected. When screening of women is involved, cost-effectiveness declines as HIV prevalence falls, because of the larger number of women who must be screened to identify an HIV-positive woman (Rely and others 2003).
Feeding substitution. Whereas in high-income countries the health community recommends complete avoidance of breastfeeding for HIV-infected mothers to prevent postnatal HIV transmission, in developing countries the feasibility of this approach is often limited by such factors as cost, sustainability, lack of safe water, health, and child spacing and by sociocultural factors (Coutsoudis 2002). Prolonged breastfeeding more than doubles the likelihood of MTCT (Nduati and others 2000). Because evidence indicates that mixed feeding (breast milk and formula or other substance) has a higher risk of transmission than exclusive breastfeeding (Coutsoudis and others 1999), mothers should be counseled on the superiority of early weaning over mixed feeding. Even fewer data are available on the cost-effectiveness of feeding substitution.
Prevention of Bloodborne Transmission
Below we discuss the effectiveness and cost-effectiveness of harm reduction for injecting drug users, implementation of blood safety practices, and provision of sterile injections:
Harm reduction for injecting drug users. Harm reduction involves a combination of health promotion strategies for users, including needle and syringe exchange programs, ready access to effective drug treatment and substitution, and provision of counseling and condoms. Brazil, which has reduced the incidence of HIV and kept HIV prevalence from reaching projected levels, has relied on strong official support for harm reduction as a cornerstone of its national prevention program (Mesquita and others 2003). A limited number of studies have shown significant reductions in HIV incidence among those exposed to needle exchange programs, and several studies have shown significant reductions in needle sharing (see table 18.3). Methadone maintenance is both safe and effective as a treatment for drug addiction (National Consensus Development Panel on Effective Medical Treatment of Opiate Addiction 1998) and may help reduce the risk of HIV transmission by enabling individuals to avoid the drug-using behaviors that can lead to HIV infection (Metzger, Navaline, and Woody 1998; Needle and others 1998). However, the effect of drug treatment modalities on the rate of HIV transmission is currently limited by laws in many countries that prohibit or restrict the use of methadone maintenance or other drug substitution strategies. The evidence supporting the cost-effectiveness of needle exchange programs in high-income countries is strong. However, little has been published in relation to developing countries, partly because these programs have not been as widely implemented as hoped. Given the low cost of syringes, the extremely high efficiency of HIV transmission by this route, and the demonstrated effectiveness of harm reduction programs in changing syringe-sharing behavior, needle exchange programs should be one of the most cost-effective interventions.
Implementation of blood safety practices. Transmission of HIV can be virtually eliminated in health care settings through a blood safety program that ensures (a) a national blood transfusion service; (b) the recruitment of voluntary, low-risk donors; (c) the screening of all donated blood for HIV; and (d) the reduction of unnecessary and inappropriate transfusions (UNAIDS 1997). Available evidence indicates that HIV screening is effective in reducing HIV infections (see table 18.4). Blood screening for HIV is costly but has been shown to be cost-effective in numerous studies in developing countries (see table 18.3) (Foster and Buve 1995; Hutton, Wyss, and N'Diekhor 2003; Watts, Goodman, and Kumaranayake 2000). The evidence appears to support the WHO and UNAIDS recommendations that all countries, regardless of the nature of the epidemic in the country, should implement a comprehensive blood safety program.
Universal precautions. A critical component of standard infection control in health care settings is a prohibition on reusing needles and syringes. A controversy has recently arisen among researchers who contend that HIV infections have been significantly misclassified because of the under-counting of cases that result from unsafe injection practices by misattributing such cases to heterosexual transmission (Gisselquist and others 2003). However, after much investigation, WHO and the U.S. Department of Health and Human Services concluded that even though transmission caused by unsafe injections may have been underreported, it nevertheless does not account for an appreciable amount of HIV transmission (WHO and UNAIDS 2003). Cost-effectiveness analyses indicate that a combined policy strategy of single-use syringes and interventions to minimize injection use could reduce injection-related infections by as much as 96.5 percent, or 8.86 million DALYs between 2000 and 2030, at an average cost of US$102 per DALY. Additional cost-effectiveness studies are needed to guide decisions regarding the optimal choice of technology in this area.
To prevent bloodborne transmission of HIV and other diseases, health care workers, emergency personnel, and others who might experience occupational exposure to blood or body fluids are advised to take universal precautions. This approach, which treats all bodily fluids as potentially infectious, includes the use of gloves, gowns, and goggles; the proper disposal of waste; and the use of sterile injection and other infection control practices (CDC 1989). Studies have demonstrated that the use of protective gear, such as gloves, reduces the likelihood of blood exposure in health care settings.
Although the cost-effectiveness of implementing universal precautions increases as HIV prevalence increases, universal precautions are unlikely to be cost-effective in resource-limited settings especially where HIV prevalence is low. Postexposure prophylaxis with antiretroviral agents is considered the standard of care after occupational needle-stick exposure to blood from an HIV-infected person. Cost-effectiveness analyses of postexposure prophylaxis have been conducted only in high-income countries and have concluded that this intervention is not cost-effective (Low-Beer and others 2000; Pinkerton, Holtgrave, and Bloom 1998).