Background Mathematical modelling of sexually transmitted infections suggests that targeting intervention (TI) to high-risk heterosexual risk groups (HRG) who have disproportionately high exposure and potential for transmission within populations can be very effective. We reviewed HIV transmission modelling studies to better understand the potential impact of TIs or the contribution of HRG to overall HIV transmission across geographical regions and epidemic phases.

Methods We systematically searched PubMed with relevant key words to identify publications that used dynamical models of heterosexual HIV transmission, and then searched papers to identify studies that incorporated heterogeneity in risk, and provided estimates of the population attributable fraction of HIV infections due to HRGs (PAF), or fraction of infections prevented (PF) or change in prevalence due to TIs.

Results Of 917 titles, 283 were excluded on abstract review. Of 634 papers searched, 96 modelled heterogeneity, of which 26 were included. Six studies used non-regionalised models, 9 studied generalised epidemics (GE) in sub-Saharan Africa, nine studied concentrated epidemics (CE) in Asia, West Africa, Japan, and Europe, and two studied both epidemic types. The PAF of HRGs ranged from 13% to 17% in mature GEs with an HIV prevalence of 16%–22% across three studies. Five models explored TIs in GEs and predicted a PF of 12%–73% and a 0%–27% reduction in prevalence with >50% coverage of commercial partnerships. Ten studies modelled TIs in CEs, with overall HIV prevalence at the mature phase between 0.7% and 3.5%, and suggested that TIs could reduce prevalence by 14%–30%, with PFs of 25%–48% if >75% coverage of commercial partnerships. With <50% coverage of commercial partnerships, 1 study demonstrated a 14% reduction in prevalence at 10 years, and two studies predicted a PF between 13% and 20%. The PF of TIs implemented early in a CE with high coverage ranged between 27% and 97%. Two studies predicted that additional TIs (pre-exposure prophylaxis) associated with high levels of risk compensation in mature epidemic settings could reverse positive gains already made by increased condom use see Abstract O1-S06.01 table 1.

Conclusion Modelling studies suggest that TIs have the potential to reduce HIV in the overall population in generalised and concentrated epidemics. The relative impact of TIs depends on coverage, epidemic phase, differential risk between HRGs and remainder of the population, and the time-scale of outcome measurement.