EMCDDA Document library

Trajectories of injecting behavior in the Amsterdam Cohort Study among drug users


Longitudinal injecting patterns were similar in the cohort to earlier US reports. Duration of injecting/year of cohort entry were associated with these patterns. Changes in injecting of drugs over time were mirrored in the incidence of HIV.

Main subject: 


This abstract is provided here as a convenience only. Check the publisher's website (if available) for the definitive version.


Injecting frequency among people who inject drugs (IDU) can change along distinct trajectories, which can reflect on incidence of HIV and HCV infections. We aimed at assessing these patterns of longitudinal changes, their predictors and their association with the incidence of HIV and HCV.


We analyzed data from the Amsterdam Cohort Study among Drug Users, selecting participants recruited from 1985 to 2005, injecting drugs before cohort entry and with records in at least three different six months intervals (N = 740). We used latent class mixed models to identify distinct trajectories of injecting, multinomial regression to identify socio-demographic variables associated with those patterns and Kaplan-Meier analysis for the estimation of the corresponding cumulative HIV and HCV incidence.


Five distinct patterns for injecting frequency and for injecting since last visit were identified. The majority of participants (three groups, 69% of participants) had stable risk injecting behavior; the remaining displayed a decrease in injecting over time. Those with longer duration of injecting at cohort entry and those who entered the cohort in earlier years tended to have continuing high risk behavior. The HIV risk was highest among those with continuing high risk behavior and its changes over time mirrored the patterns of change in injecting in a group with decrease in injecting.


Individual longitudinal patterns of changes in injecting behavior are related to socio-demographic and drug use variables and are reflected in the incidence of HIV infections. Understanding these associations might provide valuable information for targeted interventions.