SHM Converge Scientific Abstract ePoster Gallery-Targeting Goal of Care Discussion to High-Risk Patients Based on Predictive Analytics Improved Hospitalists' Code Status Documentation Compliance

Targeting Goal of Care Discussion to High-Risk Patients Based on Predictive Analytics Improved Hospitalists' Code Status Documentation Compliance

Back to course

Pdf Summary

Lehigh Valley Health Network improved hospitalists’ compliance with code status discussions and documentation by combining education with predictive analytics. Hospitalists were trained on the importance of these conversations and how to conduct them. The team then used two predictive models—the End-of-Life (EOL) Care Index and the Inpatient Deterioration Index (IDI)—to identify high-risk patients most likely to benefit from goals-of-care discussions.<br /><br />A Best Practice Advisory (BPA) alert was built into the electronic health record to prompt clinicians when a patient had no documented code status discussion and met high-risk criteria. The alert linked directly to the appropriate documentation tool and was limited to daytime hours to reduce alarm fatigue, especially for nocturnists.<br /><br />The EOL Care Index predicts one-year mortality risk, while the IDI predicts acute deterioration, ICU transfer, code blue, or mortality within 38 hours. Together, these tools allowed the team to target reminders to the patients most likely to need urgent conversations, rather than prompting for all admissions.<br /><br />As a result, code status discussion and documentation improved substantially, reaching about 70% to 80% in high-risk patients compared with a baseline of about 50% to 60% across all patients. The project also improved overall institutional code status documentation, DNR status documentation, and ICU transfer outcomes. The authors conclude that predictive analytics can efficiently support patient-centered care by focusing clinicians on the right patients while minimizing alert fatigue. Future work will use this model to expand goal-directed, patient-focused interventions.

Asset Subtitle

Shadi Jarjous

Meta Tag

Author List Shadi Jarjous, Zhe Chen

Category Innovations

Concept Predictive Analytics

Concept Code Status Discussion

Concept EOL Care Index

Concept Inpatient Deterioration Index

Concept Risk Stratification

Distinguished Non-Finalist

Presenter Organization Jefferson Health-Lehigh Valley Region

Presenting Author Shadi Jarjous

Track Technology in Hospital Medicine

Keywords

code status discussions

predictive analytics

hospitalist compliance

goals of care

electronic health record

best practice advisory

end-of-life care index

inpatient deterioration index

alarm fatigue

patient-centered care

Predictive Analytics

Code Status Discussion

EOL Care Index

Inpatient Deterioration Index

Risk Stratification