Concurrent partnerships in men do not explain HIV incidence in women: number of partners does

This article is more than 13 years old. Click here for more recent articles on this topic

A study of HIV incidence amongst women in part of KwaZulu-Natal, South Africa, failed to find any evidence that HIV incidence in women was associated with overlapping, or concurrent, relationships in their male partners.

The study, led by Dr Frank Tanser at the Africa Centre for Health and Population Studies at the University of KwaZulu-Natal, was published in The Lancet on July 15.

However, the study has confirmed that the more partners a woman’s male partner has, the higher the HIV risk is to each individual woman. This finding, plus confirmation of a strong association between the number of partners the women themselves had and their HIV infection rate, provides backing for clear and simple partner reduction advice campaigns for both men and women.

The concurrency hypothesis

This study's findings are important because one hypothesis developed over at least the last decade to explain why HIV prevalence is so much higher in southern Africa than elsewhere is that in certain African cultures there is a high prevalence and acceptability, in both men and women, of long-term, concurrent sexual relationships (Halperin 2004, and see Overlapping relationships in Preventing HIV). This is because this pattern maximises the number of people in a community who are sexually connected at any one time.

Glossary

hypothesis

A tentative explanation for an observation, phenomenon, or scientific problem. The purpose of a research study is to test whether the hypothesis is true or not.

multivariate analysis

An extension of multivariable analysis that is used to model two or more outcomes at the same time.

p-value

The result of a statistical test which tells us whether the results of a study are likely to be due to chance and would not be confirmed if the study was repeated. All p-values are between 0 and 1; the most reliable studies have p-values very close to 0. A p-value of 0.001 means that there is a 1 in 1000 probability that the results are due to chance and do not reflect a real difference. A p-value of 0.05 means there is a 1 in 20 probability that the results are due to chance. When a p-value is 0.05 or below, the result is considered to be ‘statistically significant’. Confidence intervals give similar information to p-values but are easier to interpret. 

mathematical models

A range of complex mathematical techniques which aim to simulate a sequence of likely future events, in order to estimate the impact of a health intervention or the spread of an infection.

One mathematical model back in 1997, for instance (Morris) found that when the mean number of concurrent partnerships in a population was 1.68, the largest single sexually interlinked network comprised 2% of the local population. When the mean number of concurrent partners increased to 1.86, no less than 64% of the sexually active population became linked into a single network. A real-life study on an island in Lake Malawi (Helleringer) appeared to confirm this.

More recently, however, critics of the concurrency hypothesis (Sawers, Lurie) have asserted that the evidence for an association between concurrency and HIV prevalence is weak.  

Study method

The new study in The Lancet appears to confirm this, at least as applied to rural KwaZulu-Natal, one of the highest HIV prevalence areas in the world.

The problem with establishing or disproving the concurrency hypothesis is that it is not a theory about women’s risk of acquiring HIV, but of their partners’ risk of transmitting it. It does not link sexual risks an individual takes with their risk of infection, but rather links it to multiple risks taken by their partners. A rigorous study would therefore have to measure not only the number of partners women had but the number of partners each one of their partners had, and the degree to which these partners’ relationships overlapped – clearly a huge task.

The Lancet study got round this ingeniously by using the fact that in rural South Africa most people have sexual partners who live very near them: few people see partners who live a long distance away, and, in this area, there are relatively few partners who are migrant workers.

It therefore linked men and women not by whether they actually were partners, but by whether they lived in the same location. It broke down a 434 km2 area into individual ‘pixels’ with a radius of three kilometres each (28km2 in area). It measured HIV incidence in women and the number of partners they had in each micro-area, and also measured in the same micro-area the number of partners men had and whether they had more than one sexual partnership going on at the same time (i.e. concurrency). It then worked out whether there was an association between the number and concurrency of partnerships in men with HIV incidence in women.

Incidence in women could be established because the area is subject to regular surveillance by the Africa Centre Demographic Information System. This regularly collects HIV risk information and also conducts tests in people aged over 15 in the area. An estimate of incidence can therefore be made by observing the rate at which people become HIV-positive through successive surveys.

The sexual behaviour data for men, however, were only collected once, at the start of the study in 2004. Participants were asked how many lifetime sexual partners they had and how many they were currently involved in. If they said more than one, they were regarded as being in concurrent relationships.

Findings

This is one of the highest-prevalence areas for HIV anywhere in the world. A quarter of the adult population has HIV, and prevalence peaks at 50% in women aged 25 to 30 and 44% in men aged 30 to 35. Overall incidence is 7.5 per 100 people a year and peaks at 7.5% a year in women aged 24 and 5% a year in men aged 29.

Interestingly, marriage was a minority practice in this area, with only 31% of women and 23% of men ever having been married.

Both the average number of partners men had and the average number in concurrent relationships varied hugely by micro-area. The average number of partners men had ranged from 3.4 to 12.9, and the percentage of men involved in concurrent relationships ranged from 4 to 76.3%. In the east of the area, which contained the only semi-urban locale, the township of KwaMsane, men reported high lifetime numbers of sexual partners but relatively few concurrent partnerships; conversely, in the rural west, men reported fewer lifetime sex partners but a higher proportion were in concurrent relationships.

Firstly, the study reconfirmed that there was an extremely strong relationship between more than one sexual partner in women and HIV infection. Computed annual HIV incidence in women reporting no sexual partners in the last year was 0.94% (the average gap between HIV tests was 1.8 years), in women reporting one partner was 4.5% and in women reporting more than one partner was the extremely high figure of nearly 12%.

Secondly, the study found a strong association between the number of partners men had and HIV incidence in women. HIV incidence in women was 3.16% a year in the areas inhabited by the 25% of men with the lowest number of partners, and 4.37% in the areas inhabited by the 25% of men with the highest number of partners. After multivariate analysis, it was calculated that for every increase of one in the lifetime number of partners men in each locale reported, the risk of acquiring HIV in women went up by 8%. This was statistically significant (p=0.004). Women living in areas with men reporting the highest number of partners (over twelve) had nearly double the risk of acquiring HIV compared with women living in areas with men reporting the lowest number of partners (below four).

Thirdly, however, and by contrast, there was no association seen between concurrency in men and HIV incidence in women. HIV incidence was 3.4% in women in areas where men had the lowest number of concurrent relationships and 3.5% in areas with the highest frequency of concurrency, and there was no association between male concurrency and female incidence (p=0.73).

Implications and comments

There are limitations to this study. The men were only asked about the number of partners they had at the start of the study, and this could possibly have changed during the study; and they were only asked whether they were in a concurrent relationship right now, rather than whether they tended to have them. The researchers acknowledge that their findings, featuring a mature epidemic, don’t rule out the possibility that concurrency, and therefore highly connected networks, may have a role to play in the early stages of an epidemic, simply by ensuring HIV reaches more people more quickly.

The researchers point out that their findings provide evidence for the simplest interpretation of the link between multiple partnership and HIV – people with more partners both acquire and transmit it more often – and none for the more complex idea of concurrence, Furthermore, they add, campaigns warning people against concurrency may inadvertently given the impression that “having many serially monogamous relationships does not place an individual at significant risk of infection.

“Conversely,” they add, “simplifying the public health message to reduction in multiple partnerships alone is likely to improve message clarity and effectiveness.”

In an accompanying editorial, Nancy Padian of the US Global AIDS Coordinator’s office and Shanthi Manian of the Bill and Melinda Gates Foundation say that this study should serve to inject clarity into messages aimed at young African people.

“Messages should be explicit about the behavioural change required and appropriate for context,” they say.

“Studies...suggest that young people do not understand global catchphrases such as those about faithfulness” and may interpret ‘faithfulness’ as meaning trust rather than monogamy.  This study, they add, “reinforces the need for simple, unambiguous prevention messages to discourage individuals from having several sexual partners, whether concurrent or not.”

References

Tanser F et al. Effect of concurrent sexual partnerships on rate of new HIV infections in high-prevalence rural South Africa: a cohort study. The Lancet 378:247-255, 2011.

Halperin D, Epstein H Concurrent sexual partnerships help to explain Africa’s high HIV prevalence: implications for prevention. The Lancet, 364(1):4-6, 2004.

Morris M, Kretzschmar M Concurrent sexual partnerships and the spread of HIV. AIDS 11:681-83, 1997

Helleringer S, Kohler HP The structure of sexual networks and the spread of HIV/AIDS in rural Malawi. Population Association of America, annual meeting, 2006.

Lurie MN, Rosenthal S Concurrent partnerships as a driver of the HIV epidemic in sub-Saharan Africa? The evidence is limited. AIDS Behav 14:17–24, 2010.

Padian NS and Manian S. The concurrency debate: time to put it to rest. Lancet 378:203-204, 2011.