For the second phase of study selection (Phase 2), we incorporated input from the expert consultants and further restricted our criteria. Studies were included if they were published from 2000 through 2008 to reflect more recent data. Studies were excluded if they provided inadequate description of methods or were not ongoing national, population-based surveys. We restricted to ongoing surveys to allow for multiple years of data to provide a more robust estimate of PWID. In addition, we manually searched for national household surveys that measured injection drug use and applied the Phase 2 selection criteria. Five reports identified in Phase 1 were excluded and 2 national surveys were included from the manual search for a total of 4 surveys used in meta-analysis (Figure 1).
