Columbia University (Helleringer), University of Malawi (Kalilani-Phiri), University of Pennsylvania (Frimpong), Montfort Hospital, Malawi (Mkandawire)

According to the abstract of this research done in Likoma, a small island in the northern region of Lake Malawi in the country of Malawi: "Multiple concurrent partnerships (MCP) have been described as the 'key driver' of generalized HIV epidemics, but comparative studies show that MCP are not more common in regions of Africa with high HIV prevalence than in concentrated epidemics. This might be due to systematic under-reporting of MCP during sexual behavior surveys. We use sexual network data from a small island on Lake Malawi, in conjunction with a simple model of HIV transmission, to assess whether self-reports of partnership concurrency possibly underestimate the contribution of MCP to HIV epidemics. We find that the prevalence of MCP was significantly higher in this population according to estimates derived from network data. Self-reported data underestimated the proportion of incident infections attributable to MCP by more than 50%. Our results emphasize the need for interventions addressing patterns of sexual networking at the population level rather than focusing on the transmission of HIV within stable cohabiting couples."

In this study, multiple concurrent partnership (MCP) - defined as having 2 or more sexual partnerships - was examined by using network studies based on partner tracing designs. Following a household census that enumerated all inhabitants of the island, eligible respondents were asked to provide the names of their sexual partners. They were also asked to answer questions about the context of their relationships with these partners (e.g., duration and timing), and relationship-specific risk factors for HIV transmission (e.g., condom use). "All data are linked and allow reconstructing chains of sexual relations connecting members of the population of interest. Such data may be less vulnerable to under-reporting of sexual relationships because one’s relationship(s) are potentially reported not only by the respondent him/herself, but also by his/her sexual partner(s)....The number of sexual partners a respondent reports during a sexual network survey is sometimes called the 'outdegree'.... The number of times a respondent is nominated by other survey respondents during a sexual network survey, on the other hand, is the 'indegree'. The combination of indegrees and outdegrees is referred to as the 'total degree' of a respondent. If self-reports of sexual behaviors were perfectly accurate, outdegree, indegree and total degree would be equal for each member of the network. In the presence of under-reporting however, the measures diverge. In this paper, we systematically compare inferences about the role of MCP in HIV transmission made on the basis of outdegrees alone to inferences made on the basis of total degrees."

The authors analysed data from the Likoma Network Study (LNS), which traced the sexual networks of young adults on Likoma. Interviews were conducted using audio computer-assisted self-interviewing techniques (ACASI), "which have been shown to significantly increase the validity of reports of stigmatized behaviors in other context". The network of sexual relationships was then constructed by tracing all nominated partners and linking, where possible, nominated partners to survey respondents and individuals included in the household roster (more than 80% of partners residing on the island). An HIV test of participants was also conducted using two rapid tests as suggested by the Malawi Ministry of Health.

The study's analytic strategy is available on pages 4 and 5 of the document, outlining its statistical and modelling analysis. Among other results detailed in the analysis, the study showed that self-reports of sexual relationships underestimated the prevalence of MCP in this population by 112% for women and 52.3% for men. In comparing self-reporting (SR-MCP) individuals' health to non-reporting (NR-MCP) individuals' health, "..estimates derived from network data indicate...: women in MCP were generally more likely to have used condoms with any of their current partners, but were significantly less likely to have done so consistently...[and] ...the prevalence of STIs [sexually transmitted infections] was significantly higher among women with NR-MCP relative to women with SR-MCP....Among men, the prevalence of condom use was higher among MCP than among men in serial relations according to self-reported data, as was the proportion of men worrying about HIV." The report's data show that self-reported data underestimate the impact of partnership concurrency on HIV transmission by more than 50% on average.

In conclusion, the study finds that self-reported data are not well-suited to estimate the prevalence of MCP and assess the contribution of concurrency to the spread of HIV. It recommends relying on sexual network data that include partner tracing. "First of all, self-reported data significantly underestimated the prevalence of partnership concurrency within this population. Second, several co-factors potentially enhancing the transmission of HIV during sexual intercourse were more common among respondents with NR-MCP than among the rest of the population. Third, self-reports of MCP also underestimated the proportion of incident HIV infections attributable to MCP according to a simple model of HIV transmission."

Population Association of America (PAA) website [11], accessed on October 28 2009.