Estimating the number of people who inject drugs using repeated respondent-driven sampling (RDS) in a community-based program: implications for the burden of hepatitis C and HIV infections and harm reduction coverage.

Estimates of the population size of people who inject drugs (PWID) are essential for efficient program planning and for monitoring key targets. Existing estimates in Greece are based on the capture-recapture method applied to drug treatment sources. We aimed to obtain estimates based on data collected from a community-based program addressing PWID in Athens, Greece. The program was implemented in 2012-2013 to increase diagnosis and treatment for HIV among PWID during an HIV outbreak. Five Responden-Driven Sampling (RDS) rounds were used to recruit participants. A unique code was used to identify participants among rounds. Capture-recapture was applied to estimate the population size in 2013 (PWID with injection in the past 12 months; active PWID with injection in the past 30 days). Log-linear models were applied. In 2013, the estimated number of active PWID in Athens was 4,117 [95% confidence interval (CI): 3,728-4,507] (vs. 1,956 [95% CI: 1,525-2,565] the existing population size estimate). Based on this estimate, the coverage of needle and syringe programs in 2013 was 103 syringes/PWID/year (vs. 216 based on the existing estimate). The population prevalence of injecting drug use in Athens (past 12 months) was 0.222% (95% CI: 0.200-0.245). The inclusion of data from community-based programs in the estimation of the PWID population size resulted in 2.1-fold higher estimates, compared to the official estimates obtained from drug treatment data, and indicates the need for re-evaluation of necessary resources for harm reduction and elimination of HIV and hepatitis C in PWID.