At this point we convened an expert consultation to provide feedback on best methods for producing population estimates of PWID for short-term use (i.e., based on available data) and for longer-term use, which may require primary data collection. The group of 10 experts represented persons who directed some of the 7 data systems identified in the Phase 1 automated search, those who had calculated PWID population estimates, and researchers studying PWID populations. For the short term, the expert consultants endorsed the use of meta-analysis of national surveys for making PWID population estimates, acknowledging the limitations of these surveys. National household surveys have the advantage of robust sampling methods and allow for population-based inference; however, they are less robust for measurement of injection drug use due to exclusions (people who are homeless or institutionalized), stigma and social desirability influences related to reporting injection drug use behavior, and small sample sizes of PWID. The expert consultants recommended that we assess the meta-analysis findings against other data sources that may not be included in the meta-analysis.
