Injection drug network characteristics as a predictor of injection behaviour.

Social network characteristics of people who inject drugs (PWID) have previously been flagged as potential risk factors for HCV transmission such as increased injection frequency. To understand the role of the injecting network on injection frequency, we investigated how changes in an injecting network over time can modulate injecting risk behaviour. PWID were sourced from the Networks 2 Study, a longitudinal cohort study of PWID recruited from illicit drug street markets across Melbourne, Australia. Network-related correlates of injection frequency and the change in frequency over time were analysed using adjusted Cox Proportional Hazards and Generalised Estimating Equations modelling. Two-hundred and eighteen PWID followed up for a mean (s.d.) of 2.8 (1.7) years were included in the analysis. A greater number of injecting partners, network closeness centrality and eigenvector centrality over time were associated with an increased rate of infection frequency. Every additional injection drug partner was associated with an increase in monthly injection frequency. Similarly, increased network connectivity and centrality over time was also associated with an increase in injection frequency. This study observed that baseline network measures of connectivity and centrality may be associated with changes in injection frequency and, by extension, may predict subsequent HCV transmission risk. Longitudinal changes in network position were observed to correlate with changes in injection frequency, with PWID who migrate from the densely-connected network centre out to the less-connected periphery were associated with a decreased rate of injection frequency.