Using Machine Learning and Behavioral Nudges in Electronic Health Records to Reduce Opioid Overdose Risk in Adults

What this Trial Is About

This clinical trial focuses on patients at risk of opioid overdose identified through a machine-learning based risk prediction model. The study aims to evaluate a behavioral nudge intervention delivered via the Electronic Health Record EHR to improve opioid prescribing safety and reduce overdose risk. The trial compares usual care to interventions involving an elevated-risk flag and behavioral nudges targeted at clinicians in primary care settings. Participants are divided into three groups usual care without changes, an elevated-risk flag displayed in the EHR during patient encounters, and an elevated-risk flag combined with best practice alerts or behavioral nudges that prompt clinicians under specific conditions. These nudges encourage safer prescribing behaviors such as naloxone prescription and require clinicians to justify certain opioid-related orders. During the study, participants are monitored through primary care visits where the EHR interventions are applied. Outcomes include composite scores of prescribing practices assessed at 4 and 6 months after enrollment, as well as measures such as naloxone prescriptions, opioid dosage levels, overlapping prescriptions, and emergency visits related to overdose. The researchers will evaluate changes in clinician prescribing behavior and patient safety over the course of the trial.

Machine-Learning Prediction and Reducing Overdoses With EHR Nudges