Modeling longitudinal trajectories of ketamine injection among young injecting drug users

Background: No previous studies have modeled trajectories of ketamine injection longitudinally, which may offer important insights into patterns of cessation and continuation. Methods: A sample of 68 young ketamine injecting drug users (IDUs) – aged 16 to 29 - was interviewed about ketamine injection every three to four months over a two-year period. Longitudinal models of ketamine injection were developed using Proc Traj (SAS). Trajectories of ketamine injection frequency were illustrated using a censored normal model of standardized rates of use at each follow-up. Trajectories of ketamine injection likelihood were illustrated using a logistic model of use (yes/no) at each follow-up. Findings: Approximately 50% did not inject ketamine at any follow-up. A censored normal model indicated that 42.5% followed a flat linear, low-use trajectory – injecting ketamine approximately once per year during follow-up, while 7.8% followed a declining linear, high-to-low use trajectory – decreasing ketamine injection from a mean rate of once per month to once per year. A logistic model revealed that 33.9% – nearly all injected ketamine in the year prior to first interview - had a 50% likelihood of injecting ketamine at each follow-up (declining linear trajectory) while 16.3% - none injected ketamine in the year prior to first interview – increased to a 10% likelihood of injecting ketamine at each follow up (slightly increasing linear trajectory). Co-variates, including gender and homeless status, improved model fit but were not statistically significant. Conclusions: Longitudinal models indicated that most IDUs ceased ketamine injection or injected sporadically, while frequent injectors decreased use during the study.