SHM Converge Scientific Abstract ePoster Gallery-Predictors of Post-Discharge Adverse Events for Hospitalized Patients With Multiple Chronic Conditions

Pdf Summary

This retrospective study examined 293 hospitalized adults with multiple chronic conditions discharged from general medicine at an academic medical center in Boston. Researchers aimed to build an interoperable prediction model for post-discharge adverse events (AEs) using standardized patient-reported outcomes (PROs), structured electronic health record (EHR) data, and established risk scores.<br /><br />All patients completed a 10-item global health questionnaire at discharge. Investigators used validated chart review with two-clinician adjudication to identify AEs within 14 and 30 days after discharge. Twelve candidate predictors were considered, including demographics, medication changes, pending tests, Epic’s Readmission Risk score, Epic’s Deterioration Index, and PRO-based physical and mental health scores. Multivariable logistic regression was used to evaluate associations and model performance.<br /><br />The models showed only modest ability to discriminate patients at risk for post-discharge AEs, especially within 14 days. The strongest predictors reflected clinical deterioration or hemodynamic instability. A moderately elevated Deterioration Index and low mean arterial pressure were independently associated with 14-day AEs, while low mean arterial pressure was also independently associated with 30-day AEs. Female sex and non-White race/ethnicity showed trends toward significance. Neither the Readmission Risk score nor the PRO measures were associated with adverse events.<br /><br />The study concludes that incorporating deterioration-related variables into discharge planning may improve identification of high-risk patients. Future work may use large language models to extract additional factors, such as functional or cognitive status, from clinical notes to strengthen prediction.

Asset Subtitle

Savanna Plombon

Meta Tag

Author List Anuj K. Dalal, Jorge A. Rodriguez, Kaitlyn A. Konieczny, Madeline Smith, Maria Edelen, Marie Leeson, Robert S. Rudin, Savanna Plombon, Stuart Lipsitz

Category Research

Concept Predictive Model

Concept Post-Discharge Adverse Event

Concept Patient-Reported Outcome

Concept Electronic Health Record

Concept Deterioration Index

Distinguished Non-Finalist

Presenter Organization Brigham and Women's Hospital

Presenting Author Savanna Plombon

Track Patient Safety

Keywords

post-discharge adverse events

multiple chronic conditions

predictive model

patient-reported outcomes

electronic health record

deterioration index

mean arterial pressure

logistic regression

hospital discharge

retrospective study

Predictive Model

Post-Discharge Adverse Event

Patient-Reported Outcome

Electronic Health Record

Deterioration Index