Machine-Learning Prediction and Reducing Overdoses With EHR Nudges

What this Trial Is About

Researchers are evaluating a clinician-targeted behavioral nudge intervention embedded in the Electronic Health Record (EHR) to help identify patients at higher risk for opioid overdose using a machine-learning based risk prediction model. This trial aims to improve opioid prescribing safety and reduce overdose risk by comparing usual care to enhanced care with flags and behavioral nudges for clinicians in primary care settings. The study includes three groups: one receiving usual care without changes, one with an elevated-risk flag displayed in the EHR during patient encounters, and one with the flag combined with best practice alerts or behavioral nudges triggered under specific conditions. These nudges encourage safer prescribing, such as prescribing naloxone or providing justification for high opioid doses or overlapping prescriptions. Participants are patients who have received an opioid prescription and visited a primary care practice in the last year. Researchers will assess prescribing practices at 4 and 6 months after enrollment, including measures like naloxone prescriptions, opioid dosage, and overlapping medications. The study will monitor emergency department or inpatient visits for overdose and overall visits to evaluate the intervention's impact on patient safety and prescribing behavior.

Machine-Learning Prediction and Reducing Overdoses With EHR Nudges