Columbia Model Predicts Return-to-Use Risk After Treatment for Opioid Use Disorder Interview with:
Assistant Professor of Clinical Psychiatry
Data Science Research Group
Division on Substance Use Disorders
Department of Psychiatry
Columbia University What is the background for this study?

Response: Opioid use disorder presents a major public health crisis, with increasing overdose death through the last 5 years. Treatment delivery continues to be difficult, with a large number of patients not stably maintained on Medication for Opioid Use Disorder (MOUD) after the initial treatment engagement.

In this study we applied novel statistical methods to a newly harmonized dataset incorporating 3 large clinical trials from the National Drug Treatment Clinical Trials Network (CTN) to develop individual level risk prediction models for opioid use disorder. We showed that urine toxicology data in the first 3 weeks after initiation of treatment can predict return-to-use at the end of 3 months with surprising accuracy. What should readers take away from your report?

Response: Clinicians should use the CORRS score to calculate the risk of return-to-use for patients engaged in Medication for Opioid Use Disorder (MOUD). CORRS is easy to calculate. It’s the number of positive or missing urine toxicology for non-prescribed opioid in the first 3 weeks of treatment.

Our model shows that with 0 positive or missing urine, the risk of return-to-use at the end of 3 months is 13%. With 1, it is 22%; with 2 it is 53% and with 3 it is 85%. Baseline variables can be used to further improve this prediction or make a less accurate prediction at treatment imitation. That model is disseminated as a public domain product at What recommendations do you have for future research as a result of this study?

Response: Future direction of this research goes in two main directions.

One direction is to create a real time prediction score, that is analogous to a consumer credit score, that tells you if a patient will return to the clinic or return-to-use in the next week. Accompanying that would be a opioid risk report, similar to a credit report, that provides clinicians some understanding of risk factors, some modifiable, in a clear, concise way. Our group has received a National Institute of Health research project award (R01) to develop these tools for clinicians and researchers.

A second line of future research involves replication of this predictive model in special populations, assessing its algorithmic bias and fairness, and potentially leveraging it as a way to strategy higher and lower risk patients to different interventions in future clinical trials.

No disclosures

Luo SX, Feaster DJ, Liu Y, et al. Individual-Level Risk Prediction of Return to Use During Opioid Use Disorder Treatment. JAMA Psychiatry. Published online October 04, 2023. doi:10.1001/jamapsychiatry.2023.3596

The information on is provided for educational purposes only, and is in no way intended to diagnose, cure, or treat any medical or other condition. Always seek the advice of your physician or other qualified health and ask your doctor any questions you may have regarding a medical condition. In addition to all other limitations and disclaimers in this agreement, service provider and its third party providers disclaim any liability or loss in connection with the content provided on this website.

Last Updated on October 4, 2023 by