Knowing the prevalence and correlates of multiple and concurrent sexual partnerships is important for understanding the dynamics of HIV transmission, and thus for developing effective prevention interventions. Although at least a few theoretical models of multiple and concurrent partnerships have been developed, there is little agreement about how to derive empirical measures and how to assess the relationship of multiple and concurrent sexual partnerships with HIV infection.
This study takes advantage of self-reported data on sexual partnerships and biomarker data on HIV serostatus that have been collected in recent years from adult women and men (age 15-49) by nationally representative Demographic and Health Surveys (DHS) and AIDS Indicator Surveys (AIS). Using information on up to three of the respondents’ most recent sexual partners, we evaluate and compare the prevalence of concurrent sexual partnerships across countries—defining concurrent partnerships as having two or more sexual partners that overlapped in time in the year preceding the survey. We also examine key characteristics of respondents reporting concurrent partnerships in pooled samples for sub-Saharan Africa, and we evaluate the association between concurrency and HIV serostatus at the individual level, after controlling for educational level, wealth status, condom use, male circumcision, and other factors. Finally, we assess the relationship between prevalence of concurrency and HIV prevalence at the community and country levels.
We find that men are much more likely than women to have concurrent partners. Our analysis also shows that many reported multiple partnerships in the 12 months preceding the survey interview were not concurrent ones. Finally, very few men had overlapping partners for one year or longer.
In the pooled samples for sub-Saharan Africa, we find that urban, more-educated, and wealthier women and men are more likely to have concurrent partnerships than their rural, less-educated, and poorer counterparts. Circumcised men are also more likely to have concurrent partners than uncircumcised men. Those who had concurrent partners are more likely to report using condoms than those who did not have concurrent partners; yet only one-fifth of women and less than one-tenth of men with concurrent partners reported using condoms at last sex.
In most countries, at the individual level women and men who had concurrent sexual partners in the previous 12 months were more likely to be HIV-positive than those who had only one lifetime partner, or those who had multiple lifetime partners but no overlapping partners in the previous 12 months. Yet the duration of overlap in concurrent sexual relationships does not seem correlated with the likelihood of HIV infection. At the individual level, in the pooled samples for sub-Saharan Africa, a positive and significant relationship between concurrent sexual partnerships and HIV-positive status is observed for both women (aOR=3.32; 95%CI: 2.22-4.97) and men (aOR=2.87; 95%CI: 1.85-4.45), after adjusting for other factors such as educational level, wealth status, urban/rural residence, and condom use. Among men, controlling for male circumcision has virtually no effect on the adjusted association between sexual concurrency and HIV serostatus (aOR=2.85; 95%CI: 1.84-4.42).
In multivariate models, associating one’s concurrency behavior with his/her HIV serostatus reveals that the likelihood of HIV infection is only slightly greater among individuals with concurrent partnerships in the previous 12 months (aOR=3.32 for women; aOR=2.87 for men) than among those with multiple lifetime partnerships that were not concurrent in the previous 12 months (but could have been previously) (aOR=2.86 for women; aOR=2.63 for men). This is to be expected because having concurrent partners increases the risk of transmitting HIV infection to the partners, not necessarily one’s own risk of infection above the risk of having multiple serial partners. One’s own risk may be greater only to the extent his/her concurrency behavior is a proxy for partners’ concurrency behavior or belonging to a higher-risk sexual network.
The prevalence of sexual concurrency does not seem correlated with HIV prevalence at the community level or at the country level, neither among women nor among men. The associations are even weaker when the prevalence of HIV among women is correlated with the prevalence of concurrency among men, and when the prevalence of HIV among men is correlated with the prevalence of concurrency among women. The lack of a relationship between the prevalence of concurrency and HIV prevalence among men at the community level does not seem due to varying prevalence levels of male circumcision. However, at the country level a stronger association between prevalence of concurrency among men and HIV prevalence emerges in countries with lower prevalence of male circumcision.
The study identifies a number of measurement issues and data constraints that limited the scope of our analysis and that should be kept in mind when interpreting the findings and planning future studies. Some of the major limitations of the study include the cross-sectional and self-reported nature of the survey data, the lack of data on complete sexual histories, and the lack of data on sexual networks.
Some of these data limitations have already been addressed in more recent DHS and AIS surveys by systematically including questions about the number of the respondent’s lifetime sexual partners, and about consistent condom use with all partners (up to three) in the previous 12 months. The measurement of concurrency could be further improved by collecting information on the duration of the sexual relationship with each of the respondent’s sexual partners in the previous 12 months, including his/her spousal partners, and by collecting information on the frequency of sexual intercourse during each relationship.
Despite the limitations inherent to the measurement of concurrency using self-reported data from cross-sectional population-based surveys, the findings of this study shed new light on the prevalence and correlates of concurrency, as well as on the association between concurrency and HIV.