Concurrent sexual partnerships and the spread of HIV - ‘the evidence is limited’

The theory that multiple, overlapping sexual partnerships are a key driver of generalised HIV epidemics in Africa has been attacked as being based on insubstantial evidence. The critics, writing in the journals AIDS and Behavior and The Lancet, argue that researchers lack a precise definition of concurrency or a standard way to measure it, and that the data do not show a significant association between concurrency and either HIV incidence or prevalence.

However this critique has stimulated a fierce debate in the United States. Proponents of the concurrency thesis argue that the critics’ analysis of the data is selective, that evidence from a wide range of sources supports the thesis, and that it would be irresponsible for prevention programmes in Africa to ignore this issue.

Concurrency

Concurrent sexual partnerships describe situations in which an individual has overlapping sexual relationships with more than one person. They can be contrasted with serial monogamy, when an individual has a sexual relationship with only one partner, with no overlap in time with subsequent partners.

A number of researchers, including Daniel Halperin, Timothy Mah and Martina Morris have suggested that concurrent relationships can increase the size of an HIV epidemic, the speed at which it infects a population and its persistence within a population.

The explanation for this is that in situations where a significant proportion of both men and women have concurrent relationships, even if they only have two partners each, as soon as one person in the network of concurrent relationship contracts HIV, then other people in the network are at risk (unless condoms are used). More people are more often exposed to the virus, including during the acute infection period when people are extremely infectious.

In contrast, in situations of serial monogamy, even if men and women have a relatively high number of sexual partners during their lifetime, one relationship is over before another is started. This means that if HIV is passed on within a relationship, it cannot be further transmitted as long as that relationship lasts. (For more background, click here.)

Critique

Mark Lurie and Samantha Rosenthal however argue that even if this theory is persuasive, empirical evidence for it is lacking. They call for better designed studies to clarify the contribution concurrency may make to generalised epidemics in Southern and Eastern Africa. Moreover, they believe that delivering prevention interventions around concurrency could be counter-productive and may divert resources away from other prevention methods that have proven efficacy.

Lurie and Rosenthal argue that concurrency is often vaguely and inconsistently defined. Some studies in fact collect data on total numbers of partners, and not concurrent relationships. Moreover, some include very brief or one-off liaisons (e.g. with a sex worker), but most do not.

Mah and Halperin, proponents of the concurrency thesis, accept that the lack of a consensus definition of concurrency or of a universally accepted method of measurement hampers comparison between studies.

They report a proposed standard definition from a UNAIDS working group: overlapping sexual partnerships where sexual intercourse with one partner occurs between two acts of intercourse with another partner.

Lurie and Rosenthal also point out that even studies using the same methodology have produced conflicting results.

The debate between Lurie and the other researchers focuses on the different type of research studies that may or may not demonstrate the contribution that concurrency makes.

The prevalence of concurrency in a population

A number of studies have taken a representative sample of a population to quantify the proportion of people who are participating in concurrent sexual relationships. These surveys show wide variation between different countries, with populations in sub-Saharan Africa tending to report much more concurrency than populations in other parts of the world.

Lurie and Rosenthal’s main criticism is that these studies simply cannot tell us anything about a link between concurrency and HIV.

In addition, they question the validity of comparisons between countries, given the variety of definitions used by researchers. They believe that there is no substantial evidence that levels of concurrency are significantly higher in Africa than elsewhere.

Lurie and the other authors tussle over the same studies. Referring to a review of sexual behaviour in 59 countries, Lurie insists that it found that concurrency rates could not be compared and that African adults are less sexually active than adults in other regions.

In contrast, Mah and Halperin provide the following quote from the same study: ‘‘Evidence is available that, although lifetime numbers of partners might be lower, concurrent relationships in men in some African countries might have been more common and of longer duration than in other regions’’.

Lurie and Rosenthal respond that this is a hypothesis, not a proven finding, and that more research is needed.

Qualitative data

Whereas Mah and Halperin believe that qualitative research can demonstrate that concurrency is a highly normalised behaviour in many parts of southern and eastern Africa and can help us understand its socio-cultural underpinnings, but Lurie and Rosenthal point out that qualitative studies “show that there are different types of concurrent relationships and that treating them all as if they are the same may be counter-productive.”

Individual studies

Only a few studies have compared individuals’ participation in concurrent relationships and their HIV status, and Lurie and Rosenthal note that a consistent relationship has not been found.

However Martina Morris, a leading researcher of concurrency, argues that such studies are “theoretically misguided and empirically irrelevant”. She says that concurrency is not a risk for the person who has more than one partner, but a risk for that person's partners. A monogamous partner may be exposed to HIV, not by his or her own behaviour, but by the partner’s concurrency. Because of this, future studies will need to enrol partners.

Mah and Halperin also believe that concurrency increases an individual’s risk of transmitting HIV, not their risk of acquiring it. They point to studies from Uganda and Zimbabwe where HIV infection was associated with the belief that one’s partner was having concurrent relationships.

Population studies

In 2001, Lagarde and colleagues reported a study that used a standardised questionnaire to assess concurrency rates and HIV prevalence in five sub-Saharan cities. The study did not find that the two factors were correlated - for example, some lower prevalence cities had high rates of concurrency.

Lurie and Rosenthal cite this as a key study, but Martina Morris rejects the study design entirely. This is because HIV prevalence represents infections that have accumulated over many years, whereas the survey measured concurrency only in the previous year.

Mathematical modelling

Lurie and the proponents of concurrency all agree that the most powerful demonstrations of the influence of concurrency have come from simulation models. For example, Martina Morris and Mirjam Kretzschmar worked on Ugandan data and concluded that increasing the level of concurrency would have a more significant impact on epidemic spread than increasing the number of partnerships.

Lurie and Rosenthal say that even if these models show that concurrency can drive an epidemic, such theoretical work cannot demonstrate whether concurrency is actually doing so in Africa.

They also comment that other modelling studies, which found that the total number of partners or mixing between different social groups were more important than concurrency, tend not to be cited by the other authors.

In addition, in the articles published by Lurie and Helen Epstein, there is much claim and counter-claim as to the definitions used and the validity of the assumptions that were fed into the various modelling studies.

Conclusions

Mark Lurie and Samantha Rosenthal believe that the evidence base for the role of concurrency is weak and contradictory, and that better research with more refined definitions needs to take place before interventions to reduce concurrency can be delivered.

Morris counters that the studies Lurie and Rosenthal have looked at cannot prove or disprove the hypothesis. More sophisticated studies are being worked on and will give a more precise picture of concurrency’s role, “but no one argues that concurrency is irrelevant to transmission,” she says.

As such, she says it would be a “real tragedy” if methodological limitations were used to justify a do-nothing policy.

Mah and Halperin also argue that if HIV prevention interventions were never implemented until the most reliable evidence had been gathered, the only ones in use today would be male circumcision and interventions to prevent mother-to-child transmission. They believe that prevention messages which encourage people to have only one partner at a time are needed as one component of a prevention response.