An Examination of Older Adults Seeking Rehabilitation/Residential Treatment for Opioid Use Disorder in the United States
Open Access DepositedDownloadable Content
Analysis 1
Opioid use disorder (OUD) treatment retention is critical to addressing the opioid crisis and reducing overdose deaths among older adults in the United States. This study aimed to examine the association between planned receipt of medication-assisted opioid therapy (MAOT) and treatment retention in short-term and long-term rehabilitation/residential treatment among older adults with OUD and identify key predictors of treatment retention using a machine learning approach. Four classification models (logistic regression, logistic regression with elastic net penalty, random forests, and boosted classification trees) were developed and 23 potential key predictors of treatment retention were analyzed, including sociodemographic, treatment, and substance use-related characteristics. Optimal model tuning parameters were selected using 10-fold cross-validation and the best models were selected using cross-validated average ROC-AUC, sensitivity, and specificity. Logistic regression models were used to determine the association between MAOT and treatment retention and the best performing models were used to determine the key predictors of treatment retention. We found that older adults with planned MAOT use had higher odds of treatment retention in short-term settings but lower odds of treatment retention in long-term settings. The boosted classification tree models performed the best for both residential treatment settings. Treatment referral source, census region, source of income, and frequency of primary opioid use at admission were identified as top influential predictors of treatment retention among older adults in both residential settings. Future studies should further explore these key predictors of treatment retention among older adults to inform opioid treatment programs.
Important and underappreciated age group differences among adults seeking treatment for opioid use disorder in rehabilitation/residential settings in the United States may exist. We aimed to characterize these differences and compare correlates of treatment completion by age. Using data from the 2015-2019 Treatment Episode Dataset-Discharges (TEDS-D), we described and compared sociodemographic, treatment, and substance use characteristics across different age groups (18-29, 30-49, 50-54, 55-64, and 65+) of adults seeking treatment for OUD in rehabilitation/residential settings. We used logistic regression to determine the age-stratified associations between these characteristics and treatment completion. We found several age-group differences in demographic, treatment and substance use characteristics and correlates of treatment completion. Notably, discharge cases in the 55-64 group who were married (vs never married) had higher odds of treatment completion, whereas those in the 18-29 group who were married (vs never married) had a lower odds of treatment completion. Among discharge cases in the 18-29, 30-49, and 50-54 groups, an older age at first opioid use (18-29) was associated with a higher odds of treatment completion compared to a younger age at first opioid use (under 18). Conversely, in the 55-64 group, an older age of first opioid use (30+ vs under 18) had a lower odds of treatment completion. Our results suggest that there are distinct age differences among younger and older adults seeking treatment for OUD in rehabilitation/residential settings in the US. Age-specific approaches should be developed and implemented to improve treatment for older adults. Analysis 2
Common patterns of substance use behavior among older adults seeking treatment for OUD is currently not known. We aimed to classify older adults seeking rehabilitation/residential treatment for OUD into substance use behavior classes, compare sociodemographic characteristics between classes, and evaluate the association between class membership and treatment completion. Using data from the 2015-2019 Treatment Episode Dataset-Discharges (TEDS-D), we performed a latent class analysis (LCA) to classify older adults into substance use behavioral classes and determined sociodemographic and treatment-related correlates of class membership using multinomial logistic regression. We also used logistic regression to determine the association between substance use behavioral class and treatment completion. We found four latent classes of substance use behavior among older adults seeking treatment for OUD in residential settings, 1) daily some injection heroin use with stimulant use (26.1%, n=9,403), 2) daily some injection heroin use with alcohol and stimulant use (9.9%, n=3,565), 3) daily some injection heroin use (50.4%, n=18,154), and 4) daily non-injection other opioid use (13.7%, n=4,926). Correlates of being in one of the daily some injection heroin use classes compared to the daily non-injection other opioids class included being younger, male, Non-Hispanic Black, never married, less than high school educated, unemployed, homeless, receiving public assistance or no source of income, recently arrested one or more times, treated in the Northeast region, previously treated one or more times, planned to be treated using MAOT, and younger at first use. The daily some injection heroin use, daily some injection heroin use with stimulant use, and the daily some injection heroin use with alcohol and stimulant use classes had lower odds of treatment completion compared to the daily non-injection other opioid use class. Our results suggest that there are groups of older adults with OUD that have distinct substance use behaviors, sociodemographic characteristics, and higher likelihood of treatment completion. Analysis 3
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.