Predicting Primary Care Physician Burnout From Electronic Health Record Use Measures

Area of Science:

• Health Informatics
• Medical Informatics
• Physician Well-being

Background:

• Physician burnout is a significant issue impacting healthcare quality.
• Electronic Health Records (EHRs) generate extensive usage data.
• Predicting burnout risk is crucial for timely intervention.

Purpose of the Study:

• To assess if routine EHR use metrics can predict physician burnout.
• To identify clinical work units at higher risk for burnout.
• To evaluate EHR data for targeted intervention strategies.

Main Methods:

• Observational study of primary care physicians.
• Collected clinical workload and EHR efficiency measures.
• Linked EHR data with 2 years of well-being surveys (Stanford Professional Fulfillment Index).
• Utilized gradient boosting classifier and other predictive models.
• Evaluated model performance using Area Under the Receiver Operating Characteristics Curve (AUC).

Main Results:

• 233 physicians completed 396 surveys; 28% reported high burnout.
• Gradient boosting model achieved an AUC of 0.59 for predicting burnout.
• Predictive features included physician age and note/order customization.
• Clinic-level measures showed 56% sensitivity and 85% specificity in identifying high-risk clinics.

Conclusions:

• Routinely collected EHR use measures have limited predictive power for individual physician burnout.
• These measures offer moderate ability to identify clinics with elevated burnout risk.
• Further research may refine EHR data utilization for burnout prediction and prevention.