From First Use to First Treatment: An Examination of the Path from Use Initiation to Treatment-Seeking for Heroin Users Across the U.S. over the Period 2000-2019

James Duggan

Advisor: Jonathan F Schulz, PhD, Department of Economics

Committee Members: Jonathan Beauchamp, Robert Axtell

Online Location, Zoom Link: https://gmu.zoom.us/j/92987591949?pwd=Rjrr72oataChuYZhpNbgQQh2aDgG3X.1
August 26, 2025, 03:00 PM to 04:00 PM

Abstract:

The opioid epidemic continues to be a major public health issue in the United States, imposing significant costs on individual users, their families, and society. Providing expeditious treatment is an effective way to combat the epidemic, but treatment for heroin users is often delayed for many years following use initiation. Little is known empirically about these delay periods or how they may vary across demographic groups. This work helps fill that knowledge gap by examining records of over 460,000 heroin users who sought and received treatment for the first time over the years 2000-2019. The nation-level data include males and females, either Black or White, twelve years of age or older, with treatment delay measured as the time elapsed, in years, from individuals’ first heroin use to their first treatment for drug use.

Chapter 1 introduces treatment delay conceptually and motivates its study. A brief review of economic models underlying addictive consumption and their implications for treatment delay is followed by a review of prior treatment delay studies and related studies with treatment delay implications, namely those pertaining to the general likelihood of receiving treatment though not considering delay explicitly.

Chapter 2 employs survival analysis techniques to examine treatment delay empirically in a time-to-event framework. Nonparametric Kaplan-Meier curves compare group-level treatment delay rates among race and gender groups and establish baseline relationships between delay period duration and event probability. A logistic hazard model then provides parametric estimates of event probability at the individual level, accounting for the effects of multiple covariates in addition to race and gender. From these estimates, standardized treatment delay curves are computed. The standardized curves control for the effects of covariates distributed differently by race or gender groups on these groups’ average delay curves. Results show that treatment delays for heroin users can be quite long, are longer for males than females, and substantially longer for Blacks than Whites.

Chapter 3 builds on the analysis in Chapter 2 by examining if and how covariates’ effects change over the delay period, while introducing into the analysis geographically-varying treatment accessibility measures constructed from external data sources. Treatment delay outcomes are classified into five ordered categories based on the length of delay. Outcome probabilities are examined using a generalized ordered logit model, which estimates separate sets of coefficients for explanatory variables depending on the outcome category. This allows for differentiating the magnitude, and possibly direction, of covariates’ effects between levels of the outcome variable. Results show that increasing the proportion of states’ treatment programs that accept Medicaid could shorten average treatment delays, more so for Blacks than Whites.