Skip to content

Abstract Search

Substance Use

Engagement in substance use disorder treatment after an emergency department visit among persons at high risk of opioid overdose: A prediction analysis Fiona Bhondoekhan* Fiona Bhondoekhan Yu Li Laura C. Chambers Francesca L. Beaudoin Benjamin D. Hallowell Linda Mahoney Mackenzie M. Daly Jamieson Goulet Brandon D.L. Marshall

Background: Predictors of substance use disorder (SUD) treatment engagement after opioid-related emergency department (ED) visits are uncertain. We sought to identify predictors of treatment engagement to inform linkage services in the ED setting.

Methods: This analysis used data from a randomized trial comparing the effectiveness of certified peer recovery specialist (CPRS) and licensed clinical social worker (LCSW) linkage services for ED patients at risk of opioid overdose in Rhode Island (2018-2021). Potential predictors were obtained from a baseline questionnaire. SUD treatment engagement within 90 days of ED discharge was obtained from administrative data linkage. Classification and regression tree (CART) models identified predictors of treatment engagement by linkage service. Model performance was assessed using predictive accuracy, sensitivity, specificity, and positive/negative predictive value.

Results: Of 648 participants (CPRS=323, LCSW=325), 350 (54%) were aged 31-50 years, 206 (32%) were male, and 278 (43%) engaged in SUD treatment within 90 days (CPRS=44%, LCSW=42%). In the CPRS service, predictors of treatment engagement were stable housing, no history of treatment barriers, history of treatment/recovery services, no prior bipolar disorder diagnosis, no prescription drug use in last 6 months, and current unhealthy alcohol use. In the LCSW service, predictors were having health insurance, no overdose in last 12 months, not currently in pain, and no prior mental illness treatment or anxiety disorder diagnosis. CART analyses had low predictive accuracy (CPRS=61%, LCSW=55%, Table 1), suggesting these measures had minimal utility for predicting treatment engagement.

Conclusion: For ED patients at risk of opioid overdose, measured characteristics did not adequately predict SUD treatment engagement following an ED visit involving CPRS or LCSW services. It remains important to offer behavioral/social support services to ED patients at risk of opioid overdose.