Dynamics of HIV epidemics in sub-Saharan Africa: introduction.

Abstract

Introduction For years, some scientists have argued that definition of the epidemiology of HIV was complete; modes of HIV transmission are known and risk factors for sexual transmission have been described. According to this perspective, prevention efforts should focus only on vigorously scaling-up sustainable interventions known to be effective in preventing HIV transmission. Although this argument has some validity, the question remains as to why HIV prevalence varies so greatly in different settings. Could answers to this question help us refocus on new or more appropriate interventions? Indeed, two decades into addressing this pandemic, the diverse dynamics of HIV transmission at population levels are still only partly understood. This is particularly true in sub-Saharan Africa, where HIV spreads heterosexually in adults, and where HIV prevalence in some cities has stabilized at around 30% of the adult population, showing little or no sign of decline. How can persistence at these high levels be explained, despite the impact of AIDS-related mortality depleting the most vulnerable segments of the population? For many years, the differing rates of HIV morbidity in various parts of Africa were attributed mainly to different timing of the introduction of the virus into the general population - as if the level of HIV should ineluctably, as a sort of natural evolution, reach prevalences of about 30%. However, the persistence over time of huge differences between African countries in HIV prevalence has made this explanation less and less plausible. The hypothesis that certain HIV-1 clades have greater infectivity has also been raised to explain rapidly growing regional epidemics, but this hypothesis remains unsubstantiated [1]. Finally, variations in vulnerability leading to increased HIV sexual exposure and greater efficiency of transmission [related to different patterns of sexual networking and differing prevalences of sexually transmitted diseases (STDs), for example], were believed to be major epidemiological factors explaining more generalized HIV epidemics in some regions. The multicentre study published in this supplement was carried out in four African cities, specifically to help understand why HIV has spread at different rates and reached different prevalences in these cities, and to better define the most important facilitating factors that could be the target of appropriate interventions. This study represents the first historical attempt to directly address these issues, at a population level, in a systematic and standardized way. For operational and practical reasons, the study was limited to four cities, Kisumu, Ndola, Yaounde and Cotonou (two with high HIV prevalence and two with relatively low HIV prevalence), which precludes generalization and most formal statistical ecological analyses. The cross-sectional study design limits inferences of cause-and-effect relationships. Nonetheless, standardized techniques of data collection included enumeration of sex workers and face-to-face interviews, and standardized laboratory procedures were used. The concurrent surveys of general population and sex-worker samples represent unique strengths of the study, as does the use of a sexual network module that collected unusually detailed information on up to the last eight non-spousal partnerships. Major STDs were assessed, as was the distribution of circulating HIV-1 subtypes. Surprisingly, with the exception of young age at first marriage, young age of women at sexual debut, and large age differences between spouses, most other parameters of risky sexual behaviour (such as contact with sex workers, lifetime number of sexual partners, rate of acquisition of new partners and lack of condom use) were not consistently more common in the high HIV prevalence sites than in the relatively low prevalence sites. Similarly, age differences between sexual partners and the frequency of concurrent partnerships were not consistently higher in Kisumu and Ndola than in Yaounde and Cotonou. The possibility remains that important culturally-driven behaviours were not measured, or that changes in behaviour that occurred over time were not measured (e.g., deaths of the most vulnerable because of high-risk behaviours who were the first to acquire HIV would have been more common in the high prevalence cities, resulting in measurement of lower prevalences of risk-taking in the surviving members of the population). In addition, differences in rates of gonorrhoea, chlamydial infection, and syphilis did not explain the differences in HIV prevalence in the four cities. However, bacterial vaginosis, which has been associated with increased risk of HIV acquisition and has been highly prevalent in East African women, was not assessed. Also, the prevalence of vaginal trichomoniasis was, in fact, higher in the cities with highest HIV prevalence, and was also significantly associated with individual risk of HIV. In multivariate analysis, the only two key variables implicated as independent individual-level risk factors for HIV and also found to differ between low and high HIV prevalence areas were herpes simplex virus type 2 (HSV-2) seropositivity and the lack of male circumcision. The authors concluded that the efficiency of HIV transmission as mediated by these biological factors outweighed differences in sexual mixing patterns in explaining the variations in HIV prevalence between the four cities. This study is the first to examine and demonstrate at population levels the prominent role of factors that appear to enhance HIV transmission in explaining high HIV prevalence. Previous studies of female-to-male transmission of HIV in the developing world have drawn attention to the importance of such risk factors at the individual level. As early as 1989, Cameron et al.[2] suggested "differences in the aetiology and prevalence of genital ulcer disease and circumcision practices ... could explain the regional variations of HIV-1 in Africa". In Thailand, Mastro et al.[3] also reported, in 1994, that "It is noteworthy that circumcision is very uncommon in Thailand and that genital ulcer disease (largely due to H. ducreyi) is common in young Thai men". These studies on female-to-male HIV transmission focused exclusively on commercial sex contacts in single sites. This multicentre study in Africa not only extends the level of comparison to the ecologic level, but also specifically shifts attention to the young population, and to the evidence that much if not most of the HIV transmission in the high HIV prevalence settings is due to the combined effects of two factors: lack of male circumcision and high prevalence of HSV-2. In these four cities, commercial sex relationships and patterns of sexual networks around clients of sex workers were not identified as independent risk factors for HIV. This suggests that, in high HIV prevalence areas, "the high risk approach" to prevention is not sufficient; prevention efforts should also concentrate on reducing unprotected sex in young people initially embarking on sexual relations. Early marriage appears as a risk factor for both HIV and HSV-2 seropositivity, perhaps reflecting relatively intense exposure through repeated sexual intercourse, or high levels of susceptibility during sexual debut. This study also gives insights into possible explanations for why the HIV prevalence is much higher in young women than young men, as previously also shown in several population-based studies carried out in Africa [4]. The striking gender difference in age-specific HIV prevalence in Kisumu and Ndola clearly seems to point to greater biological susceptibility of young women than of young men both to HSV-2 and to HIV infection, rather than to more frequent sexual exposure of young women. In attempting to identify determinants of differences in HIV prevalence across cities, the study is fraught with many methodological difficulties and biases. Retrospective data on sexual networking suffer from social desirability and memory biases. In particular, the extent to which young women accurately report any sexual experience, age at first sex, number of partners, and older sexual partners remain questionable. In the cross-sectional measurement of long-lived infections like HIV and HSV-2 infections, behaviour change could have occurred over time so the current situation may no longer reflect the conditions in the early years of the HIV epidemic or even at the time many of these infections were acquired by study participants. Collinearity between most variables, together with modest sample sizes, often make controlling for confounding difficult. Despite these limitations, in the absence of consistent condom use, the suggested combined effect of lack of male circumcision and HSV-2 infection in favouring HIV acquisition, and the synergistic feedback between HSV-2 and HIV in increasing susceptibility to and infectivity of each other, offer plausible explanations for sustained, generalized HIV epidemics in Africa. The authors of each of the papers in this supplement briefly discuss the many important programmatic and policy implications of their findings for HIV prevention. Among these, the key implications include the following. First, the target of preventive interventions that seek to forestall transmission of HIV must extend, if not shift, from adults to young people. To reduce risk of exposure of young people to individuals with HSV-2 and/or HIV infections, targeted behavioural interventions such as community and school-based peer education should be considerably strengthened for and with young women and young men before they start sexual activity. The earliest sexual relations should be considered as a targeted priority in these high HIV prevalence settings, because the incidence of HSV-2 and HIV infections is especially high in young women after sexual debut, and after only a few sexual acts. Furthermore, it is easier to establish patterns of safer sex practices from the start than to change risk behaviours [5], and delaying first sex may also positively influence risk behaviours later in life [6]. In view of the very high susceptibility of young women to HIV infection, even modest improvements in consistent condom use and in STD case management among young people should have great effects on the spread of HIV at the population level. Two recent qualitative studies regarding first sexual encounters in the South African region have shown the critical role of gender norms and peer pressure, and the absence of any discussion of STD/HIV, pregnancy or protection [7,8]. Finally, experiences in other African countries have shown that young people who have participated in education programs on HIV prevention have great potential to adopt and sustain safer sex behaviours. Second, these findings add weight to the growing opinion of many experts that expanded access to safe male circumcision in Africa represents an HIV prevention priority. While randomized, controlled trials are needed to confirm the efficacy and safety of this approach for prevention of HIV acquisition, it is not unreasonable to proceed to gather other experience with this approach in carefully supervised and evaluated clinical feasibility studies. Male circumcision consistently has been associated with lower rates of HIV infection in sub-Saharan Africa [9]. Whether male circumcision can represent a routine component of a package of effective HIV preventive measures where male circumcision is not currently practised needs to be urgently systematically assessed by communities themselves in different cultural settings. An operational research agenda has recently been developed that should lead to feasibility studies [10]. Finally, the alarming extent of HSV-2 infection among young people and the potential role of genital herpes for enhancing HIV transmission suggested by this study call for urgent attention. The landmark study in Mwanza, Tanzania, [11] showed, after 2 years, approximately 40% lower HIV incidence in communities with improved syndromic management of symptomatic STDs as compared with communities without this intervention. This finding raised the hope that efforts to prevent HIV transmission in Africa would be strengthened to include vigorous expansion of syndromic management of STDs, including appropriate counselling. Anecdotal reports from the region indeed seem to indicate a decrease in genital ulcers caused by syphilis and chancroid, but with a relative or absolute increase in genital ulcers attributable to genital herpes. Community-based education on HSV-2 symptom recognition, temporary abstinence or condom use during outbreaks of genital herpetic lesions should be promoted. A recent international workshop in London strongly recommended further control measures such as episodic or suppressive antiviral therapy, and further development and testing of therapeutic and prophylactic HSV-2 vaccines [12]. In public health, adherence to a preventive intervention and the proportion of the population actually reached by an intervention are as important as the intrinsic efficacy of the intervention itself [13]. For communicable diseases like HIV infection, reaching those most at risk for transmitting or acquiring infection is also critical [13]. The unacceptably low rates of condom use, even in populations with high rates of HIV infection, illustrate problems with adherence to this intervention and perhaps low levels of actually reaching populations at risk with effective intervention messages. This calls for urgent action. Two decades into the HIV epidemic, this study provides national and international policy-makers and national programme managers with new potential targets to control HIV transmission in highly affected areas. Potential new targets highlighted by this multicentre study include male circumcision, prevention of HSV-2 transmission, and early detection and treatment of trichomoniasis. Furthermore, without renewed efforts on a large scale to engage new generations into safe sexual relations and increased access to appropriate STDs services, there is little hope that the natural evolution of HIV epidemics can be modified to avoid continuing high incidence and sustained high prevalence. Early detection and treatment of other STDs, and behavioural interventions designed to prevent unprotected sexual intercourse in many settings, especially in youth, are also strongly supported by data from individual risk factor analyses in this study.