Is the HCV-HIV co-infection prevalence amongst injecting drug users a marker for the level of sexual and injection related HIV transmission?

Background: Amongst injecting drug users (IDUs), HIV is transmitted sexually and parenterally, but HCV is transmitted primarily parenterally. We assess and model the antibody prevalence of HCV amongst HIV-infected IDUs (denoted as HCV-HIV co-infection prevalence) and consider whether it proxies the degree of sexual HIV transmission amongst IDUs. Methods: HIV, HCV and HCV-HIV co-infection prevalence data amongst IDU was reviewed. An HIV/HCV transmission model was adapted. Multivariate model uncertainty analyses determined whether the model's ability to replicate observed data trends required the inclusion of sexual HIV transmission. The correlation between the model's HCV-HIV co-infection prevalence and estimated proportion of HIV infections due to injecting was evaluated. Results: The median HCV-HIV co-infection prevalence (prevalence of HCV amongst HIV-infected IDUs) was 90% across 195 estimates from 43 countries. High HCV-HIV co-infection prevalences (>80%) occur in most (75%) settings, but can be lower in settings with low HIV prevalence (<10%) or high HIV/HCV prevalence ratios (HIV prevalence divided by HCV prevalence > 0.75). The model without sexual HIV transmission reproduced some data trends but could not reproduce any epidemics with high HIV/HCV prevalence ratios (>0.85) or low HCV-HIV co-infection prevalence (<60%) when HIV prevalence > 10%. The model with sexual HIV transmission reproduced data trends more closely. The proportion of HIV infections due to injecting correlated with HCV-HIV co-infection prevalence; suggesting that up to 80/60/<20% of HIV infections could be sexually transmitted in settings with HCV-HIV co-infection prevalence between 50-60/70-80/>90%. Conclusion: Substantial sexual HIV transmission may occur in many IDU populations; HCV-HIV co-infection prevalence could signify its importance.