Allocation of resources to public health responses depends on having plausible estimates of the size of the population at risk. Unfortunately, the numbers of people who inject drugs (PWID) are difficult to estimate since injection drug use is highly stigmatized. Though estimation methods exist, the robustness of the methods to their assumptions is not well understood. Comparisons between methods are also lacking; information regarding the successive-sampling method is particularly scarce. The present study used several methods—including the successive-sampling method—to produce population size estimates from three rounds of cross-sectional surveys of PWID in San Francisco. It compares these estimates across time and across method. Our summary estimates are 10,158 for 2005, 15,554 for 2009, and 22,500 for 2012. Though the point summaries suggest an increasing population, considerable uncertainty is involved. Comparisons between and within methods reveal high variability, suggesting dependence on assumptions and analytic choices. We conclude that further research is needed to improve upon the estimation methods or develop entirely new ones. Meanwhile, plausible estimates can be achieved via multiple methods while avoiding the pitfall of relying on a single method that may be highly biased and highly imprecise.
All Science Journal Classification (ASJC) codes
- Social Psychology
- Public Health, Environmental and Occupational Health
- Infectious Diseases
- Population size
- Population size estimation methods
- Respondent-driven sampling