HIV transmission in Africa: high risk groups equally important in early and advanced epidemics

Multiple sex partners, sex for pay, and sexual coinfections (particularly genital herpes, or HSV-2) continue to act as major risk factors for HIV transmission in Africa, according to a systematic review of 68 separate epidemiological studies conducted over the past 20 years. The analysis, published in the October 2007 issue of PLoS One, found that these factors have remained significant over time and have not declined in importance as HIV prevalence becomes higher in the general population.

Studies have clearly identified sexual risk factors for heterosexual transmission of HIV in Africa, including multiple sex partners, transactional or paid sex, lack of circumcision in males, and co-infection with sexually transmitted infections, notably HSV-2.

However, many epidemiologists argue that these risk factors are most important in the early stages of an epidemic's progress through a local population, when HIV still tends to be concentrated among high-risk groups (sex workers and those with higher numbers of sex partners).

According to this view (referred to by the authors of this paper as the "conventional wisdom"), these sexual risk factors contribute less to the spread of HIV as time progresses and infection becomes more common in the wider population – a model which would affect the behaviours and populations targeted by prevention programmes.

To study the role of sexual risk factors in varying populations, the authors systematically reviewed epidemiologic studies from the past two decades. The objectives were to assess the level of risk these factors play in HIV transmission, whether their impact has changed over time, and whether it has differed between populations with higher or lower prevalence of HIV infection. The study was conducted by investigators from the International Studies of HIV/AIDS project of the Centre for Global Health Research (http://www.cghr.org/), specifically established in 2003 to conduct large scale epidemiological studies in developing countries.

The team searched for all related papers published in French or English between 1986 and June 2006. Of 110 papers that initially met the study criteria, a total of 68 studies from 18 countries were included in the review, yielding a total population of 17,000 HIV-seropositive adults and 73,000 HIV-negative controls.

Most of the studies (62) were cross-sectional; six were longitudinal. The authors note the limitations presented by cross-sectional studies, which cannot ascertain cause-and-effect relationships, and note that they could not explore all other risk factors of interest. Male circumcision was not investigated, as previous studies have shown a strong protective effect that has changed little over time. For this analysis, "paid sex" was defined as (for women) ever having been a paid sex worker, and (for men) ever having been a client of a female sex worker: i.e., paid male sex work (which has been very little studied) was not considered.

Overall, the researchers found that multi-partner sex (three or more partners vs. two or fewer), paid sex (being a female sex worker or male client), STIs and HSV-2 infection consistently posed as much of a risk for HIV transmission in advanced as in early HIV epidemics. There were no differences between early and late time periods in the strength of the association between these factors and HIV infection; tests for trends toward change with time all showed non-significant results.

The researchers then analysed variations in sexual risk according to the HIV prevalence in the overall population at the time of the individual studies. Prevalence rates were assigned according to UNAIDS data from the study year or the year closest to it, stratified by urban or rural area. Rates were classified as "higher" (>20%) or "lower" (≤20%); an unpublished analysis using a 10% cutoff yielded "very similar" results.

The odds ratios for paid sex showed no differences between areas with higher or lower HIV prevalence. For women, there were no significant differences due to number of partners, history of STIs, or HSV-2 infection between areas with higher or lower HIV prevalence. For men, a greater number of partners and a history of STIs tended to show greater risk in areas with higher HIV prevalence, whereas HSV-2 infection tended toward a greater risk in areas with lower prevalence. However, none of these variations was statistically significant. (See summary table.)

Risk for HIV infection by sexual risk factors

(Odds ratio, [95% confidence interval])

*no differences by background HIV prevalence (numbers not shown).

In conclusion, the authors note that a very heterogeneous set of individual studies were reviewed, but that the primary objective was to identify "trends in risks and differences between high and low prevalence [areas]" – differences which were not observed in their analysis. They conclude that the "key sexual risk factors [examined in this overview] appear to have changed little … during the last two decades" and that their consistency over time and between populations "emphasizes the importance of high-risk sexual behaviour in transmission."

Currently, UNAIDS defines a "generalised epidemic" as one in which HIV prevalence is more than 1% in the adult general population. Other studies and commentaries have suggested that such a definition may obscure important epidemiological characteristics. The authors of the current paper agree that "there is a need for a careful re-examination of the UNAIDS definition of 'generalized' epidemic" and that prevention programmes "should focus on interventions among vulnerable groups irrespective of the population levels of infection."