Machine-Learning Prediction and Reducing Overdoses With EHR Nudges

In response to the opioid overdose crisis, health systems have instituted multiple interventions to reduce patient risk, including decreasing unsafe opioid prescribing among high-risk patients and dispensing naloxone. However, these interventions face two key challenges. First, there are limited and poorly performing tools to identify who is truly at risk of overdose, leading to burdensome interventions targeting an overly broad population or missing key high-risk individuals. Second, even with more accurate identification of high-risk patients, highly effective strategies to change clinician behavior remain limited. Common cognitive biases may underlie clinicians' lack of response to risk factors for overdose.

This project aims to address both of these limitations by combining more accurate risk prediction tools to identify those at elevated risk of opioid overdose with novel "nudge" interventions based on principles of behavioral economics that have been shown to address cognitive biases and change prescribing behavior. The primary hypothesis is that high-risk patients in primary care practices randomized to the elevated-risk flag + nudge intervention will have safer prescribing compared to usual care.