Sociology, Department of


Date of this Version



Journals of Gerontology, Series B: Psychological Sciences and Social Sciences 78:12 (2023), pp. 2111–2121.

doi: 10.1093/geronb/gbad146


Copyright © 2023 Tse-Chuan Yang, Seulki Kim, Stephen A. Matthews, and Carla Shoff. Published by Oxford University Press on behalf of the Gerontological Society of America. Used by permission.


Objectives: Recent research has investigated the factors associated with the prevalence of opioid use disorder (OUD) among older adults (65+), which has rapidly increased in the past decade. However, little is known about the relationship between social vulnerability and the prevalence of OUD, and even less about whether the correlates of the prevalence of OUD vary across the social vulnerability spectrum. This study aims to fill these gaps. Methods: We assemble a county-level data set in the contiguous United States (U.S.) by merging 2021 Medicare claims with the CDC’s social vulnerability index and other covariates. Using the total number of older beneficiaries with OUD as the dependent variable and the total number of older beneficiaries as the offset, we implement a series of nested negative binomial regression models and then analyze by social vulnerability quartiles. Results: Higher social vulnerability is associated with higher prevalence of OUD in U.S. counties. This association cannot be fully explained by the differences in the characteristics of older Medicare beneficiaries (e.g., average age) and/or other social conditions (e.g., social capital) across counties. Moreover, the group comparison tests indicate correlates of the prevalence of OUD vary across social vulnerability quartiles in that the average number of mental disorders is positively related to OUD prevalence in the least and the most vulnerable counties and social capital benefits the less vulnerable counties. Discussion: A perspective drawing upon contextual factors, especially social vulnerability, may be more effective in reducing OUD among older adults in U.S. counties than a one-size-fits-all approach.