Temporal changes in risk factors associated with HIV seroconversion among injection drug users in eastern central Canada.

OBJECTIVES

To investigate temporal trends in HIV incidence rates and to assess changes over time in associated risk factors.

METHODS

Since 1995, the SurvUDI network has conducted surveillance among IDUs recruited in harm reduction programmes in eastern central Canada. Among the 11,731 participants, 2903 repeaters were initially HIV-negative. HIV incidence was calculated and compared for two time periods (1995-2002 vs. 2003-2009). Multivariate Cox proportional hazard models with time-dependent covariates were used to assess risk factors associated with HIV seroconversion. Interactions between covariates and time periods were examined.

RESULTS

The overall HIV incidence rate was 2.7 per 100 person-years [95% confidence interval (CI) 2.4-3.1]. It significantly decreased from 3.1 per 100 person-years in 1995-2002 to 2.2 person-years in 2003-2009. Sex, needle borrowing, and cocaine as most often injected drug were independent and stable determinants of HIV seroconversion. Age, daily injection, sex work and being recruited in an urban area showed significant interactions with time. Being aged 25 years and older, injecting daily and being recruited in an urban area predicted HIV incidence in 1995-2002 but were no longer risk factors in 2003-2009. HIV incidence increased significantly among younger IDUs and sex work emerged as a new determinant of HIV incidence in 2003-2009.

CONCLUSION

HIV incidence has decreased over time but remains high among IDUs in eastern central Canada. Associations between risk factors and HIV incidence have changed. Further research is needed to better understand HIV transmission among younger IDUs and IDU sex workers.