A STANDARDIZATION MODEL FOR ESTIMATING POPULATIONS OF PERSONS WHO INJECT DRUGS IN US STATES, 2017


Author: Rosenberg E, Rosenthal E, Hall E, Udo T, Bradley H

Theme: Epidemiology & Public Health Research Year: 2019

Background: Persons who inject drugs (PWID) are a priority population for viral hepatitis and HIV
transmission in the US. Estimates of PWID population sizes, crucial for calculations of epidemiologic
rates and resource allocation, are not available for most states. We explored the feasibility of
systematically estimating PWID populations in states using national PWID prevalence data together
with state-level narcotic overdose mortality data. .
Methods: We first estimated the nationally-representative weighted population of past-year nonmedical injection drug use among 18-64 year olds based on the 2017 US National Survey on Drug
Use and Health. Using the 2017 National Vital Statistics System mortality microdata file, we then
calculated a ratio of state-level narcotic overdose mortality rates over the national average to
allocate the national PWID estimate to states. Because states with increased fentanyl in the drug
supply may have higher fatality per PWID, a sensitivity analysis used imputed 2017 mortality rates
based on linear trends from 2011-2014 mortality rates.
Results: This primary method yielded state-level estimates for the number of PWID ranging from
553 (South Dakota) to 78,805 (Pennsylvania), with Florida, Ohio, Pennsylvania ≥ 70,000. The percent
of PWID ranged from 0.11% (South Dakota) to 1.31% (West Virginia), with Ohio, Pennsylvania, West
Virginia ≥ 1.00%. In the sensitivity analysis, the range was narrower, with maximum number of PWID
71,187 (Ohio) and percent of PWID 1.05% (West Virginia).
Conclusion: This analysis provides a feasible approach for the first systematic estimates of the
percent and number of PWID in all U.S. states utilizing the most recent national survey and local
mortality data available. These estimates are likely underestimated due to limitations of the national
estimate, which is based on household surveys. As national estimates and local indicators of PWID
improve in the US, they can be synthesized via this method to provide updated estimates.
Disclosure of Interest Statement: None to declare

Download abstract Download Poster