The epidemiological and control implications of HIV transmission probabilities within partnerships.

Abstract

The spread of HIV is in part determined by the probability of transmission within a sexual partnership. The goal of the study was to examine the relationship between the per-partnership and per-act transmission probability and explore how different assumptions influence the measurement of cofactors and interventions. We defined the mathematical relationship between the transmission probability of HIV per sex act and per partnership for the proposed biologic mechanisms. For completeness we included the original assumption of a per-partnership transmission probability and also the simple binomial model, which cannot be ruled out completely because of potential measurement error in discordant partner studies. A constant per-act or per-partnership transmission probability provides unsatisfactory models of the observed relationship between the number of sex acts and the likelihood of transmission. Either there is extreme heterogeneity in the transmission likelihood between partnerships or the transmission likelihood within a partnership decreases over time. These models cause the relative risk for a partnership in which a cofactor STD is present to decrease more rapidly than would be expected. The transmission probability per-partnership is substantially reduced only when there is a dramatic reduction in unprotected acts (e.g., condom use) or the transmission probability per act (e.g., due to antiretrovirals or STD treatment). Combining interventions can sometimes generate a more-than-additive impact, particularly with extreme heterogeneity. More empirical studies are needed to develop realistic models of transmission providing quantitative understanding of the HIV transmission process.