New Decision Analytical Model Predicts Risk of Opioid Use Disorder Return

A newly developed prediction model for opioid treatment may help detect risk of opioids by 3 weeks.

Investigators, led by Sean Luo, MD PhD, of Columbia University, wanted to create a prediction model to see if people with opioid use disorder were at a risk for opioid return during early stages of treatment—which they were.¹

“Our model could be broadly used in many clinical settings as part of a clinical decision support system in an electronic health record,” investigators wrote. “A clinician who surveyed the drug use pattern at the 1-month follow-up appointment would receive, as the electronic health record output, a numeric risk estimate and can then adjust treatment according to clinical capacity and feasibility.”

For their predictive model, the team examined other research provided by the CTN dissemination library, including the studies START, POATS, X:BOT, and the corporation EMMES.

Luo and colleagues conducted data analysis from October 2019 - November 2022. Their participant sample included 2199 adult trial participants aged >18 years old. The mean age was 35.3 years, and there were 33.1% of women (n = 728).

Just 7.8% of the participants self-reported as African American (n = 175); 14.7% reported as Hispanic (n = 14.7); 70.4% reported as White (n = 1653), and 14.7% reported as other races and ethnicities.

The model was based on 4 predicators at treatment entry, which were heroin use days; morphine and cocaine positive urine drug sample (UDS) results; and heroin injection in the past 30 days.

Ultimately, adding UDS in the first 3 treatment weeks improved model performance (AUROC, 0.82; 95% CI, 0.78 - 0.85).

The investigators also calculated the CTN-0094 OUD Return-to-Use Risk Score). The team found the 3-week mark measured the risk for opioid return to use. Participants who had 3 negative urine tests had a low risk of return to use (13%) but those who scored a positive test for all 3 tests had a high risk (85%). The prediction models had an AUROC score of 0.82 (95% CI, 0.78 - 0.85) and the team wrote how it was “considerably better than chance.”

The investigators discovered that while then medications methadone, buprenorphine and naltrexone (XR-NTX) are effective, many patients returned to opioid use during treatment.

“One interesting finding was that the difference in predicted risk of return to use between 0 and 1 opioid-positive UDS result was smaller than the differences between 1 and 2,” the team wrote. “Therefore, the threshold for applying additional intervention could be set above 1 positive UDS result.”

Previous data suggest African American and other minority populations were linked to a higher risk of opioid return, but this study did not explore it in depth enough as the LASSO procedure did not select this has a predicator.

The investigators also noted computational studies showed that dosing strategies can improve results. Another strategy is to switch to an extended-release injection formulation. The team noted how methadone doses cannot be increased too quickly for safety reasons, but future studies could consider a higher target dose.

The prediction model can ultimately help clinicians in the longer run.

“Adding to existing knowledge, our results quantify this association by indicating how likely a patient is to return to opioid use,” the team wrote. “This finding might be universal: Individuals across different treatment settings, whether predominantly using prescription opioids or heroin and who were randomized to any medication, would have low return to-use risk if they are able to sustain abstinence early.”

References

1. ^{Luo SX, Feaster DJ, Liu Y, et al. Individual-Level Risk Prediction of Return to Use During Opioid Use Disorder Treatment [published online ahead of print, 2023 Oct 4]. JAMA Psychiatry. 2023;10.1001/jamapsychiatry.2023.3596. doi:10.1001/jamapsychiatry.2023.3596}